Dependency Analyzer

Package CloudForest implements ensembles of decision trees for machine learning in pure Go (golang to search engines). It allows for a number of related algorithms for classification, regression, feature selection and structure analysis on heterogeneous numerical/categorical data with missing values. These include: Breiman and Cutler's Random Forest for Classification and Regression Adaptive Boosting (AdaBoost) Classification Gradiant Boosting Tree Regression Entropy and Cost driven classification L1 regression Feature selection with artificial contrasts Proximity and model structure analysis Roughly balanced bagging for unbalanced classification The API hasn't stabilized yet and may change rapidly. Tests and benchmarks have been performed only on embargoed data sets and can not yet be released. Library Documentation is in code and can be viewed with godoc or live at: http://godoc.org/github.com/ryanbressler/CloudForest Documentation of command line utilities and file formats can be found in README.md, which can be viewed fromated on github: http://github.com/ryanbressler/CloudForest Pull requests and bug reports are welcome. CloudForest was created by Ryan Bressler and is being developed in the Shumelivich Lab at the Institute for Systems Biology for use on genomic/biomedical data with partial support from The Cancer Genome Atlas and the Inova Translational Medicine Institute. CloudForest is intended to provide fast, comprehensible building blocks that can be used to implement ensembles of decision trees. CloudForest is written in Go to allow a data scientist to develop and scale new models and analysis quickly instead of having to modify complex legacy code. Data structures and file formats are chosen with use in multi threaded and cluster environments in mind. Go's support for function types is used to provide a interface to run code as data is percolated through a tree. This method is flexible enough that it can extend the tree being analyzed. Growing a decision tree using Breiman and Cutler's method can be done in an anonymous function/closure passed to a tree's root node's Recurse method: This allows a researcher to include whatever additional analysis they need (importance scores, proximity etc) in tree growth. The same Recurse method can also be used to analyze existing forests to tabulate scores or extract structure. Utilities like leafcount and errorrate use this method to tabulate data about the tree in collection objects. Decision tree's are grown with the goal of reducing "Impurity" which is usually defined as Gini Impurity for categorical targets or mean squared error for numerical targets. CloudForest grows trees against the Target interface which allows for alternative definitions of impurity. CloudForest includes several alternative targets: Additional targets can be stacked on top of these target to add boosting functionality: Repeatedly splitting the data and searching for the best split at each node of a decision tree are the most computationally intensive parts of decision tree learning and CloudForest includes optimized code to perform these tasks. Go's slices are used extensively in CloudForest to make it simple to interact with optimized code. Many previous implementations of Random Forest have avoided reallocation by reordering data in place and keeping track of start and end indexes. In go, slices pointing at the same underlying arrays make this sort of optimization transparent. For example a function like: can return left and right slices that point to the same underlying array as the original slice of cases but these slices should not have their values changed. Functions used while searching for the best split also accepts pointers to reusable slices and structs to maximize speed by keeping memory allocations to a minimum. BestSplitAllocs contains pointers to these items and its use can be seen in functions like: For categorical predictors, BestSplit will also attempt to intelligently choose between 4 different implementations depending on user input and the number of categories. These include exhaustive, random, and iterative searches for the best combination of categories implemented with bitwise operations against int and big.Int. See BestCatSplit, BestCatSplitIter, BestCatSplitBig and BestCatSplitIterBig. All numerical predictors are handled by BestNumSplit which relies on go's sorting package. Training a Random forest is an inherently parallel process and CloudForest is designed to allow parallel implementations that can tackle large problems while keeping memory usage low by writing and using data structures directly to/from disk. Trees can be grown in separate go routines. The growforest utility provides an example of this that uses go routines and channels to grow trees in parallel and write trees to disk as the are finished by the "worker" go routines. The few summary statistics like mean impurity decrease per feature (importance) can be calculated using thread safe data structures like RunningMean. Trees can also be grown on separate machines. The .sf stochastic forest format allows several small forests to be combined by concatenation and the ForestReader and ForestWriter structs allow these forests to be accessed tree by tree (or even node by node) from disk. For data sets that are too big to fit in memory on a single machine Tree.Grow and FeatureMatrix.BestSplitter can be reimplemented to load candidate features from disk, distributed database etc. By default cloud forest uses a fast heuristic for missing values. When proposing a split on a feature with missing data the missing cases are removed and the impurity value is corrected to use three way impurity which reduces the bias towards features with lots of missing data: Missing values in the target variable are left out of impurity calculations. This provided generally good results at a fraction of the computational costs of imputing data. Optionally, feature.ImputeMissing or featurematrixImputeMissing can be called before forest growth to impute missing values to the feature mean/mode which Brieman [2] suggests as a fast method for imputing values. This forest could also be analyzed for proximity (using leafcount or tree.GetLeaves) to do the more accurate proximity weighted imputation Brieman describes. Experimental support is provided for 3 way splitting which splits missing cases onto a third branch. [2] This has so far yielded mixed results in testing. At some point in the future support may be added for local imputing of missing values during tree growth as described in [3] [1] http://www.stat.berkeley.edu/~breiman/RandomForests/cc_home.htm#missing1 [2] https://code.google.com/p/rf-ace/ [3] http://projecteuclid.org/DPubS?verb=Display&version=1.0&service=UI&handle=euclid.aoas/1223908043&page=record In CloudForest data is stored using the FeatureMatrix struct which contains Features. The Feature struct implements storage and methods for both categorical and numerical data and calculations of impurity etc and the search for the best split. The Target interface abstracts the methods of Feature that are needed for a feature to be predictable. This allows for the implementation of alternative types of regression and classification. Trees are built from Nodes and Splitters and stored within a Forest. Tree has a Grow implements Brieman and Cutler's method (see extract above) for growing a tree. A GrowForest method is also provided that implements the rest of the method including sampling cases but it may be faster to grow the forest to disk as in the growforest utility. Prediction and Voting is done using Tree.Vote and CatBallotBox and NumBallotBox which implement the VoteTallyer interface.

Package grpcreplay supports the capture and replay of gRPC calls. Its main goal is to improve testing. Once you capture the calls of a test that runs against a real service, you have an "automatic mock" that can be replayed against the same test, yielding a unit test that is fast and flake-free. To record a sequence of gRPC calls to a file, create a Recorder and pass its DialOptions to grpc.Dial: It is essential to close the Recorder when the interaction is finished. There is also a NewRecorderWriter function for capturing to an arbitrary io.Writer. To replay a captured file, create a Replayer and ask it for a (fake) connection. We don't actually have to dial a server. (Since we're reading the file and not writing it, we don't have to be as careful about the error returned from Close). A test might use random or time-sensitive values, for instance to create unique resources for isolation from other tests. The test therefore has initial values, such as the current time, or a random seed, that differ from run to run. You must record this initial state and re-establish it on replay. To record the initial state, serialize it into a []byte and pass it as the second argument to NewRecorder: On replay, get the bytes from Replayer.Initial: Recorders and replayers have support for running callbacks before messages are written to or read from the replay file. A Recorder has a BeforeFunc that can modify a request or response before it is written to the replay file. The actual RPCs sent to the service during recording remain unaltered; only what is saved in the replay file can be changed. A Replayer has a BeforeFunc that can modify a request before it is sent for matching. Example uses for these callbacks include customized logging, or scrubbing data before RPCs are written to the replay file. If requests are modified by the callbacks during recording, it is important to perform the same modifications to the requests when replaying, or RPC matching on replay will fail. A common way to analyze and modify the various messages is to use a type switch. A nondeterministic program may invoke RPCs in a different order each time it is run. The order in which RPCs are called during recording may differ from the order during replay. The replayer matches incoming to recorded requests by method name and request contents, so nondeterminism is only a concern for identical requests that result in different responses. A nondeterministic program whose behavior differs depending on the order of such RPCs probably has a race condition: since both the recorded sequence of RPCs and the sequence during replay are valid orderings, the program should behave the same under both. The same is not true of streaming RPCs. The replayer matches streams only by method name, since it has no other information at the time the stream is opened. Two streams with the same method name that are started concurrently may replay in the wrong order. Besides the differences in replay mentioned above, other differences may cause issues for some programs. We list them here. The Replayer delivers a response to an RPC immediately, without waiting for other incoming RPCs. This can violate causality. For example, in a Pub/Sub program where one goroutine publishes and another subscribes, during replay the Subscribe call may finish before the Publish call begins. For streaming RPCs, the Replayer delivers the result of Send and Recv calls in the order they were recorded. No attempt is made to match message contents. At present, this package does not record or replay stream headers and trailers, or the result of the CloseSend method.

A dynamic and extensible music library organizer Demlo is a music library organizer. It can encode, fix case, change folder hierarchy according to tags or file properties, tag from an online database, copy covers while ignoring duplicates or those below a quality threshold, and much more. It makes it possible to manage your libraries uniformly and dynamically. You can write your own rules to fit your needs best. Demlo aims at being as lightweight and portable as possible. Its major runtime dependency is the transcoder FFmpeg. The scripts are written in Lua for portability and speed while allowing virtually unlimited extensibility. Usage: For usage options, see: First Demlo creates a list of all input files. When a folder is specified, all files matching the extensions from the 'extensions' variable will be appended to the list. Identical files are appended only once. Next all files get analyzed: - The audio file details (tags, stream properties, format properties, etc.) are stored into the 'input' variable. The 'output' variable gets its default values from 'input', or from an index file if specified from command-line. If no index has been specified and if an attached cuesheet is found, all cuesheet details are appended accordingly. Cuesheet tags override stream tags, which override format tags. Finally, still without index, tags can be retrieved from Internet if the command-line option is set. - If a prescript has been specified, it gets executed. It makes it possible to adjust the input values and global variables before running the other scripts. - The scripts, if any, get executed in the lexicographic order of their basename. The 'output' variable is transformed accordingly. Scripts may contain rules such as defining a new file name, new tags, new encoding properties, etc. You can use conditions on input values to set the output properties, which makes it virtually possible to process a full music library in one single run. - If a postscript has been specified, it gets executed. It makes it possible to adjust the output of the script for the current run only. - Demlo makes some last-minute tweaking if need be: it adjusts the bitrate, the path, the encoding parameters, and so on. - A preview of changes is displayed. - When applying changes, the covers get copied if required and the audio file gets processed: tags are modified as specified, the file is re-encoded if required, and the output is written to the appropriate folder. When destination already exists, the 'exist' action is executed. The program's default behaviour can be changed from the user configuration file. (See the 'Files' section for a template.) Most command-line flags default value can be changed. The configuration file is loaded on startup, before parsing the command-line options. Review the default value of the CLI flags with 'demlo -h'. If you wish to use no configuration file, set the environment variable DEMLORC to ".". Scripts can contain any safe Lua code. Some functions like 'os.execute' are not available for security reasons. It is not possible to print to the standard output/error unless running in debug mode and using the 'debug' function. See the 'sandbox.go' file for a list of allowed functions and variables. Lua patterns are replaced by Go regexps. See https://github.com/google/re2/wiki/Syntax. Scripts have no requirements at all. However, to be useful, they should set values of the 'output' table detailed in the 'Variables' section. You can use the full power of the Lua to set the variables dynamically. For instance: 'input' and 'output' are both accessible from any script. All default functions and variables (excluding 'output') are reset on every script call to enforce consistency. Local variables are lost from one script call to another. Global variables are preserved. Use this feature to pass data like options or new functions. 'output' structure consistency is guaranteed at the start of every script. Demlo will only extract the fields with the right type as described in the 'Variables' section. Warning: Do not abuse of global variables, especially when processing non-fixed size data (e.g. tables). Data could grow big and slow down the program. By default, when the destination exists, Demlo will append a suffix to the output destination. This behaviour can be changed from the 'exist' action specified by the user. Demlo comes with a few default actions. The 'exist' action works just like scripts with the following differences: - Any change to 'output.path' will be skipped. - An additional variable is accessible from the action: 'existinfo' holds the file details of the existing files in the same fashion as 'input'. This allows for comparing the input file and the existing destination. The writing rules can be tweaked the following way: Word of caution: overwriting breaks Demlo's rule of not altering existing files. It can lead to undesired results if the overwritten file is also part of the (yet to be processed) input. The overwrite capability can be useful when syncing music libraries however. The user scripts should be generic. Therefore they may not properly handle some uncommon input values. Tweak the input with temporary overrides from command-line. The prescript and postscript defined on command-line will let you run arbitrary code that is run before and after all other scripts, respectively. Use global variables to transfer data and parameters along. If the prescript and postscript end up being too long, consider writing a demlo script. You can also define shell aliases or use wrapper scripts as convenience. The 'input' table describes the file: Bitrate is in bits per seconds (bps). That is, for 320 kbps you would specify The 'time' is the modification time of the file. It holds the sec seconds and nsec nanoseconds since January 1, 1970 UTC. The entry 'streams' and 'format' are as returned by It gives access to most metadata that FFmpeg can return. For instance, to get the duration of the track in seconds, query the variable 'input.format.duration'. Since there may be more than one stream (covers, other data), the first audio stream is assumed to be the music stream. For convenience, the index of the music stream is stored in 'audioindex'. The tags returned by FFmpeg are found in streams, format and in the cuesheet. To make tag queries easier, all tags are stored in the 'tags' table, with the following precedence: You can remove a tag by setting it to 'nil' or the empty string. This is equivalent, except that 'nil' saves some memory during the process. The 'output' table describes the transformation to apply to the file: The 'parameters' array holds the CLI parameters passed to FFmpeg. It can be anything supported by FFmpeg, although this variable is supposed to hold encoding information. See the 'Examples' section. The 'embeddedcovers', 'externalcovers' and 'onlinecover' variables are detailed in the 'Covers' section. The 'write' variable is covered in the 'Existing destination' section. The 'rmsrc' variable is a boolean: when true, Demlo removes the source file after processing. This can speed up the process when not re-encoding. This option is ignored for multi-track files. For convenience, the following shortcuts are provided: Demlo provides some non-standard Lua functions to ease scripting. Display a message on stderr if debug mode is on. Return lowercase string without non-alphanumeric characters nor leading zeros. Return the relation coefficient of the two input strings. The result is a float in 0.0...1.0, 0.0 means no relation at all, 1.0 means identical strings. A format is a container in FFmpeg's terminology. 'output.parameters' contains CLI flags passed to FFmpeg. They are meant to set the stream codec, the bitrate, etc. If 'output.parameters' is {'-c:a', 'copy'} and the format is identical, then taglib will be used instead of FFmpeg. Use this rule from a (post)script to disable encoding by setting the same format and the copy parameters. This speeds up the process. The official scripts are usually very smart at guessing the right values. They might make mistakes however. If you are unsure, you can (and you are advised to) preview the results before proceeding. The 'diff' preview is printed to stderr. A JSON preview of the changes is printed to stdout if stdout is redirected. The initial values of the 'output' table can be completed with tags fetched from the MusicBrainz database. Audio files are fingerprinted for the queries, so even with initially wrong file names and tags, the right values should still be retrieved. The front album cover can also be retrieved. Proxy parameters will be fetched automatically from the 'http_proxy' and 'https_proxy' environment variables. As this process requires network access it can be quite slow. Nevertheless, Demlo is specifically optimized for albums, so that network queries are used for only one track per album, when possible. Some tracks can be released on different albums: Demlo tries to guess it from the tags, but if the tags are wrong there is no way to know which one it is. There is a case where the selection can be controlled: let's assume we have tracks A, B and C from the same album Z. A and B were also released in album Y, whereas C was release in Z only. Tags for A will be checked online; let's assume it gets tagged to album Y. B will use A details, so album Y too. Then C does not match neither A's nor B's album, so another online query will be made and it will be tagged to album Z. This is slow and does not yield the expected result. Now let's call Tags for C will be queried online, and C will be tagged to Z. Then both A and B will match album Z so they will be tagged using C details, which is the desired result. Conclusion: when using online tagging, the first argument should be the lesser known track of the album. Demlo can set the output variables according to the values set in a text file before calling the script. The input values are ignored as well as online tagging, but it is still possible to access the input table from scripts. This 'index' file is formatted in JSON. It corresponds to what Demlo outputs when printing the JSON preview. This is valid JSON except for the missing beginning and the missing end. It makes it possible to concatenate and to append to existing index files. Demlo will automatically complete the missing parts so that it becomes valid JSON. The index file is useful when you want to edit tags manually: You can redirect the output to a file, edit the content manually with your favorite text editor, then run Demlo again with the index as argument. See the 'Examples' section. This feature can also be used to interface Demlo with other programs. Demlo can manage embedded covers as well as external covers. External covers are queried from files matching known extensions in the file's folder. Embedded covers are queried from static video streams in the file. Covers are accessed from The embedded covers are indexed numerically by order of appearance in the streams. The first cover will be at index 1 and so on. This is not necessarily the index of the stream. 'inputcover' is the following structure: 'format' is the picture format. FFmpeg makes a distinction between format and codec, but it is not useful for covers. The name of the format is specified by Demlo, not by FFmpeg. Hence the 'jpeg' name, instead of 'mjpeg' as FFmpeg puts it. 'width' and 'height' hold the size in pixels. 'checksum' can be used to identify files uniquely. For performance reasons, only a partial checksum is performed. This variable is typically used for skipping duplicates. Cover transformations are specified in 'outputcover' has the following structure: The format is specified by FFmpeg this time. See the comments on 'format' for 'inputcover'. 'parameters' is used in the same fashion as 'output.parameters'. User configuration: This must be a Lua file. See the 'demlorc' file provided with this package for an exhaustive list of options. Folder containing the official scripts: User script folder: Create this folder and add your own scripts inside. This folder takes precedence over the system folder, so scripts with the same name will be found in the user folder first. The following examples will not proceed unless the '-p' command-line option is true. Important: you _must_ use single quotes for the runtime Lua command to prevent expansion. Inside the Lua code, use double quotes for strings and escape single quotes. Show default options: Preview changes made by the default scripts: Use 'alternate' script if found in user or system script folder (user folder first): Add the Lua file to the list of scripts. This feature is convenient if you want to write scripts that are too complex to fit on the command-line, but not generic enough to fit the user or system script folders. Remove all script from the list, then add '30-case' and '60-path' scripts. Note that '30-case' will be run before '60-path'. Do not use any script but '60-path'. The file content is unchanged and the file is renamed to a dynamically computed destination. Demlo performs an instant rename if destination is on the same device. Otherwise it copies the file and removes the source. Use the default scripts (if set in configuration file), but do not re-encode: Set 'artist' to the value of 'composer', and 'title' to be preceded by the new value of 'artist', then apply the default script. Do not re-encode. Order in runtime script matters. Mind the double quotes. Set track number to first number in input file name: Use the default scripts but keep original value for the 'artist' tag: 1) Preview default scripts transformation and save it to an index. 2) Edit file to fix any potential mistake. 3) Run Demlo over the same files using the index information only. Same as above but generate output filename according to the custom '61-rename' script. The numeric prefix is important: it ensures that '61-rename' will be run after all the default tag related scripts and after '60-path'. Otherwise, if a change in tags would occur later on, it would not affect the renaming script. Retrieve tags from Internet: Same as above but for a whole album, and saving the result to an index: Only download the cover for the album corresponding to the track. Use 'rmsrc' to avoid duplicating the audio file. Change tags inplace with entries from MusicBrainz: Set tags to titlecase while casing AC-DC correctly: To easily switch between formats from command-line, create one script per format (see 50-encoding.lua), e.g. ogg.lua and flac.lua. Then Add support for non-default formats from CLI: Overwrite existing destination if input is newer: ffmpeg(1), ffprobe(1), http://www.lua.org/pil/contents.html

A dynamic and extensible music library organizer Demlo is a music library organizer. It can encode, fix case, change folder hierarchy according to tags or file properties, tag from an online database, copy covers while ignoring duplicates or those below a quality threshold, and much more. It makes it possible to manage your libraries uniformly and dynamically. You can write your own rules to fit your needs best. Demlo aims at being as lightweight and portable as possible. Its major runtime dependency is the transcoder FFmpeg. The scripts are written in Lua for portability and speed while allowing virtually unlimited extensibility. Usage: For usage options, see: First Demlo creates a list of all input files. When a folder is specified, all files matching the extensions from the 'extensions' variable will be appended to the list. Identical files are appended only once. Next all files get analyzed: - The audio file details (tags, stream properties, format properties, etc.) are stored into the 'input' variable. The 'output' variable gets its default values from 'input', or from an index file if specified from command-line. If no index has been specified and if an attached cuesheet is found, all cuesheet details are appended accordingly. Cuesheet tags override stream tags, which override format tags. Finally, still without index, tags can be retrieved from Internet if the command-line option is set. - If a prescript has been specified, it gets executed. It makes it possible to adjust the input values and global variables before running the other scripts. - The scripts, if any, get executed in the lexicographic order of their basename. The 'output' variable is transformed accordingly. Scripts may contain rules such as defining a new file name, new tags, new encoding properties, etc. You can use conditions on input values to set the output properties, which makes it virtually possible to process a full music library in one single run. - If a postscript has been specified, it gets executed. It makes it possible to adjust the output of the script for the current run only. - Demlo makes some last-minute tweaking if need be: it adjusts the bitrate, the path, the encoding parameters, and so on. - A preview of changes is displayed. - When applying changes, the covers get copied if required and the audio file gets processed: tags are modified as specified, the file is re-encoded if required, and the output is written to the appropriate folder. When destination already exists, the 'exist' action is executed. The program's default behaviour can be changed from the user configuration file. (See the 'Files' section for a template.) Most command-line flags default value can be changed. The configuration file is loaded on startup, before parsing the command-line options. Review the default value of the CLI flags with 'demlo -h'. If you wish to use no configuration file, set the environment variable DEMLORC to ".". Scripts can contain any safe Lua code. Some functions like 'os.execute' are not available for security reasons. It is not possible to print to the standard output/error unless running in debug mode and using the 'debug' function. See the 'sandbox.go' file for a list of allowed functions and variables. Lua patterns are replaced by Go regexps. See https://github.com/google/re2/wiki/Syntax. Scripts have no requirements at all. However, to be useful, they should set values of the 'output' table detailed in the 'Variables' section. You can use the full power of the Lua to set the variables dynamically. For instance: 'input' and 'output' are both accessible from any script. All default functions and variables (excluding 'output') are reset on every script call to enforce consistency. Local variables are lost from one script call to another. Global variables are preserved. Use this feature to pass data like options or new functions. 'output' structure consistency is guaranteed at the start of every script. Demlo will only extract the fields with the right type as described in the 'Variables' section. Warning: Do not abuse of global variables, especially when processing non-fixed size data (e.g. tables). Data could grow big and slow down the program. By default, when the destination exists, Demlo will append a suffix to the output destination. This behaviour can be changed from the 'exist' action specified by the user. Demlo comes with a few default actions. The 'exist' action works just like scripts with the following differences: - Any change to 'output.path' will be skipped. - An additional variable is accessible from the action: 'existinfo' holds the file details of the existing files in the same fashion as 'input'. This allows for comparing the input file and the existing destination. The writing rules can be tweaked the following way: Word of caution: overwriting breaks Demlo's rule of not altering existing files. It can lead to undesired results if the overwritten file is also part of the (yet to be processed) input. The overwrite capability can be useful when syncing music libraries however. The user scripts should be generic. Therefore they may not properly handle some uncommon input values. Tweak the input with temporary overrides from command-line. The prescript and postscript defined on command-line will let you run arbitrary code that is run before and after all other scripts, respectively. Use global variables to transfer data and parameters along. If the prescript and postscript end up being too long, consider writing a demlo script. You can also define shell aliases or use wrapper scripts as convenience. The 'input' table describes the file: Bitrate is in bits per seconds (bps). That is, for 320 kbps you would specify The 'time' is the modification time of the file. It holds the sec seconds and nsec nanoseconds since January 1, 1970 UTC. The entry 'streams' and 'format' are as returned by It gives access to most metadata that FFmpeg can return. For instance, to get the duration of the track in seconds, query the variable 'input.format.duration'. Since there may be more than one stream (covers, other data), the first audio stream is assumed to be the music stream. For convenience, the index of the music stream is stored in 'audioindex'. The tags returned by FFmpeg are found in streams, format and in the cuesheet. To make tag queries easier, all tags are stored in the 'tags' table, with the following precedence: You can remove a tag by setting it to 'nil' or the empty string. This is equivalent, except that 'nil' saves some memory during the process. The 'output' table describes the transformation to apply to the file: The 'parameters' array holds the CLI parameters passed to FFmpeg. It can be anything supported by FFmpeg, although this variable is supposed to hold encoding information. See the 'Examples' section. The 'embeddedcovers', 'externalcovers' and 'onlinecover' variables are detailed in the 'Covers' section. The 'write' variable is covered in the 'Existing destination' section. The 'rmsrc' variable is a boolean: when true, Demlo removes the source file after processing. This can speed up the process when not re-encoding. This option is ignored for multi-track files. For convenience, the following shortcuts are provided: Demlo provides some non-standard Lua functions to ease scripting. Display a message on stderr if debug mode is on. Return lowercase string without non-alphanumeric characters nor leading zeros. Return the relation coefficient of the two input strings. The result is a float in 0.0...1.0, 0.0 means no relation at all, 1.0 means identical strings. A format is a container in FFmpeg's terminology. 'output.parameters' contains CLI flags passed to FFmpeg. They are meant to set the stream codec, the bitrate, etc. If 'output.parameters' is {'-c:a', 'copy'} and the format is identical, then taglib will be used instead of FFmpeg. Use this rule from a (post)script to disable encoding by setting the same format and the copy parameters. This speeds up the process. The official scripts are usually very smart at guessing the right values. They might make mistakes however. If you are unsure, you can (and you are advised to) preview the results before proceeding. The 'diff' preview is printed to stderr. A JSON preview of the changes is printed to stdout if stdout is redirected. The initial values of the 'output' table can be completed with tags fetched from the MusicBrainz database. Audio files are fingerprinted for the queries, so even with initially wrong file names and tags, the right values should still be retrieved. The front album cover can also be retrieved. Proxy parameters will be fetched automatically from the 'http_proxy' and 'https_proxy' environment variables. As this process requires network access it can be quite slow. Nevertheless, Demlo is specifically optimized for albums, so that network queries are used for only one track per album, when possible. Some tracks can be released on different albums: Demlo tries to guess it from the tags, but if the tags are wrong there is no way to know which one it is. There is a case where the selection can be controlled: let's assume we have tracks A, B and C from the same album Z. A and B were also released in album Y, whereas C was release in Z only. Tags for A will be checked online; let's assume it gets tagged to album Y. B will use A details, so album Y too. Then C does not match neither A's nor B's album, so another online query will be made and it will be tagged to album Z. This is slow and does not yield the expected result. Now let's call Tags for C will be queried online, and C will be tagged to Z. Then both A and B will match album Z so they will be tagged using C details, which is the desired result. Conclusion: when using online tagging, the first argument should be the lesser known track of the album. Demlo can set the output variables according to the values set in a text file before calling the script. The input values are ignored as well as online tagging, but it is still possible to access the input table from scripts. This 'index' file is formatted in JSON. It corresponds to what Demlo outputs when printing the JSON preview. This is valid JSON except for the missing beginning and the missing end. It makes it possible to concatenate and to append to existing index files. Demlo will automatically complete the missing parts so that it becomes valid JSON. The index file is useful when you want to edit tags manually: You can redirect the output to a file, edit the content manually with your favorite text editor, then run Demlo again with the index as argument. See the 'Examples' section. This feature can also be used to interface Demlo with other programs. Demlo can manage embedded covers as well as external covers. External covers are queried from files matching known extensions in the file's folder. Embedded covers are queried from static video streams in the file. Covers are accessed from The embedded covers are indexed numerically by order of appearance in the streams. The first cover will be at index 1 and so on. This is not necessarily the index of the stream. 'inputcover' is the following structure: 'format' is the picture format. FFmpeg makes a distinction between format and codec, but it is not useful for covers. The name of the format is specified by Demlo, not by FFmpeg. Hence the 'jpeg' name, instead of 'mjpeg' as FFmpeg puts it. 'width' and 'height' hold the size in pixels. 'checksum' can be used to identify files uniquely. For performance reasons, only a partial checksum is performed. This variable is typically used for skipping duplicates. Cover transformations are specified in 'outputcover' has the following structure: The format is specified by FFmpeg this time. See the comments on 'format' for 'inputcover'. 'parameters' is used in the same fashion as 'output.parameters'. User configuration: This must be a Lua file. See the 'demlorc' file provided with this package for an exhaustive list of options. Folder containing the official scripts: User script folder: Create this folder and add your own scripts inside. This folder takes precedence over the system folder, so scripts with the same name will be found in the user folder first. The following examples will not proceed unless the '-p' command-line option is true. Important: you _must_ use single quotes for the runtime Lua command to prevent expansion. Inside the Lua code, use double quotes for strings and escape single quotes. Show default options: Preview changes made by the default scripts: Use 'alternate' script if found in user or system script folder (user folder first): Add the Lua file to the list of scripts. This feature is convenient if you want to write scripts that are too complex to fit on the command-line, but not generic enough to fit the user or system script folders. Remove all script from the list, then add '30-case' and '60-path' scripts. Note that '30-case' will be run before '60-path'. Do not use any script but '60-path'. The file content is unchanged and the file is renamed to a dynamically computed destination. Demlo performs an instant rename if destination is on the same device. Otherwise it copies the file and removes the source. Use the default scripts (if set in configuration file), but do not re-encode: Set 'artist' to the value of 'composer', and 'title' to be preceded by the new value of 'artist', then apply the default script. Do not re-encode. Order in runtime script matters. Mind the double quotes. Set track number to first number in input file name: Use the default scripts but keep original value for the 'artist' tag: 1) Preview default scripts transformation and save it to an index. 2) Edit file to fix any potential mistake. 3) Run Demlo over the same files using the index information only. Same as above but generate output filename according to the custom '61-rename' script. The numeric prefix is important: it ensures that '61-rename' will be run after all the default tag related scripts and after '60-path'. Otherwise, if a change in tags would occur later on, it would not affect the renaming script. Retrieve tags from Internet: Same as above but for a whole album, and saving the result to an index: Only download the cover for the album corresponding to the track. Use 'rmsrc' to avoid duplicating the audio file. Change tags inplace with entries from MusicBrainz: Set tags to titlecase while casing AC-DC correctly: To easily switch between formats from command-line, create one script per format (see 50-encoding.lua), e.g. ogg.lua and flac.lua. Then Add support for non-default formats from CLI: Overwrite existing destination if input is newer: ffmpeg(1), ffprobe(1), http://www.lua.org/pil/contents.html

Package grpcreplay supports the capture and replay of gRPC calls. Its main goal is to improve testing. Once you capture the calls of a test that runs against a real service, you have an "automatic mock" that can be replayed against the same test, yielding a unit test that is fast and flake-free. To record a sequence of gRPC calls to a file, create a Recorder and pass its DialOptions to grpc.Dial: It is essential to close the Recorder when the interaction is finished. There is also a NewRecorderWriter function for capturing to an arbitrary io.Writer. To replay a captured file, create a Replayer and ask it for a (fake) connection. We don't actually have to dial a server. (Since we're reading the file and not writing it, we don't have to be as careful about the error returned from Close). A test might use random or time-sensitive values, for instance to create unique resources for isolation from other tests. The test therefore has initial values, such as the current time, or a random seed, that differ from run to run. You must record this initial state and re-establish it on replay. To record the initial state, serialize it into a []byte and pass it as the second argument to NewRecorder: On replay, get the bytes from Replayer.Initial: Recorders and replayers have support for running callbacks before messages are written to or read from the replay file. A Recorder has a BeforeFunc that can modify a request or response before it is written to the replay file. The actual RPCs sent to the service during recording remain unaltered; only what is saved in the replay file can be changed. A Replayer has a BeforeFunc that can modify a request before it is sent for matching. Example uses for these callbacks include customized logging, or scrubbing data before RPCs are written to the replay file. If requests are modified by the callbacks during recording, it is important to perform the same modifications to the requests when replaying, or RPC matching on replay will fail. A common way to analyze and modify the various messages is to use a type switch. A nondeterministic program may invoke RPCs in a different order each time it is run. The order in which RPCs are called during recording may differ from the order during replay. The replayer matches incoming to recorded requests by method name and request contents, so nondeterminism is only a concern for identical requests that result in different responses. A nondeterministic program whose behavior differs depending on the order of such RPCs probably has a race condition: since both the recorded sequence of RPCs and the sequence during replay are valid orderings, the program should behave the same under both. The same is not true of streaming RPCs. The replayer matches streams only by method name, since it has no other information at the time the stream is opened. Two streams with the same method name that are started concurrently may replay in the wrong order. Besides the differences in replay mentioned above, other differences may cause issues for some programs. We list them here. The Replayer delivers a response to an RPC immediately, without waiting for other incoming RPCs. This can violate causality. For example, in a Pub/Sub program where one goroutine publishes and another subscribes, during replay the Subscribe call may finish before the Publish call begins. For streaming RPCs, the Replayer delivers the result of Send and Recv calls in the order they were recorded. No attempt is made to match message contents. At present, this package does not record or replay stream headers and trailers, or the result of the CloseSend method.

Package CloudForest implements ensembles of decision trees for machine learning in pure Go (golang to search engines). It allows for a number of related algorithms for classification, regression, feature selection and structure analysis on heterogeneous numerical/categorical data with missing values. These include: Breiman and Cutler's Random Forest for Classification and Regression Adaptive Boosting (AdaBoost) Classification Gradiant Boosting Tree Regression Entropy and Cost driven classification L1 regression Feature selection with artificial contrasts Proximity and model structure analysis Roughly balanced bagging for unbalanced classification The API hasn't stabilized yet and may change rapidly. Tests and benchmarks have been performed only on embargoed data sets and can not yet be released. Library Documentation is in code and can be viewed with godoc or live at: http://godoc.org/github.com/ryanbressler/CloudForest Documentation of command line utilities and file formats can be found in README.md, which can be viewed fromated on github: http://github.com/ryanbressler/CloudForest Pull requests and bug reports are welcome. CloudForest was created by Ryan Bressler and is being developed in the Shumelivich Lab at the Institute for Systems Biology for use on genomic/biomedical data with partial support from The Cancer Genome Atlas and the Inova Translational Medicine Institute. CloudForest is intended to provide fast, comprehensible building blocks that can be used to implement ensembles of decision trees. CloudForest is written in Go to allow a data scientist to develop and scale new models and analysis quickly instead of having to modify complex legacy code. Data structures and file formats are chosen with use in multi threaded and cluster environments in mind. Go's support for function types is used to provide a interface to run code as data is percolated through a tree. This method is flexible enough that it can extend the tree being analyzed. Growing a decision tree using Breiman and Cutler's method can be done in an anonymous function/closure passed to a tree's root node's Recurse method: This allows a researcher to include whatever additional analysis they need (importance scores, proximity etc) in tree growth. The same Recurse method can also be used to analyze existing forests to tabulate scores or extract structure. Utilities like leafcount and errorrate use this method to tabulate data about the tree in collection objects. Decision tree's are grown with the goal of reducing "Impurity" which is usually defined as Gini Impurity for categorical targets or mean squared error for numerical targets. CloudForest grows trees against the Target interface which allows for alternative definitions of impurity. CloudForest includes several alternative targets: Additional targets can be stacked on top of these target to add boosting functionality: Repeatedly splitting the data and searching for the best split at each node of a decision tree are the most computationally intensive parts of decision tree learning and CloudForest includes optimized code to perform these tasks. Go's slices are used extensively in CloudForest to make it simple to interact with optimized code. Many previous implementations of Random Forest have avoided reallocation by reordering data in place and keeping track of start and end indexes. In go, slices pointing at the same underlying arrays make this sort of optimization transparent. For example a function like: can return left and right slices that point to the same underlying array as the original slice of cases but these slices should not have their values changed. Functions used while searching for the best split also accepts pointers to reusable slices and structs to maximize speed by keeping memory allocations to a minimum. BestSplitAllocs contains pointers to these items and its use can be seen in functions like: For categorical predictors, BestSplit will also attempt to intelligently choose between 4 different implementations depending on user input and the number of categories. These include exhaustive, random, and iterative searches for the best combination of categories implemented with bitwise operations against int and big.Int. See BestCatSplit, BestCatSplitIter, BestCatSplitBig and BestCatSplitIterBig. All numerical predictors are handled by BestNumSplit which relies on go's sorting package. Training a Random forest is an inherently parallel process and CloudForest is designed to allow parallel implementations that can tackle large problems while keeping memory usage low by writing and using data structures directly to/from disk. Trees can be grown in separate go routines. The growforest utility provides an example of this that uses go routines and channels to grow trees in parallel and write trees to disk as the are finished by the "worker" go routines. The few summary statistics like mean impurity decrease per feature (importance) can be calculated using thread safe data structures like RunningMean. Trees can also be grown on separate machines. The .sf stochastic forest format allows several small forests to be combined by concatenation and the ForestReader and ForestWriter structs allow these forests to be accessed tree by tree (or even node by node) from disk. For data sets that are too big to fit in memory on a single machine Tree.Grow and FeatureMatrix.BestSplitter can be reimplemented to load candidate features from disk, distributed database etc. By default cloud forest uses a fast heuristic for missing values. When proposing a split on a feature with missing data the missing cases are removed and the impurity value is corrected to use three way impurity which reduces the bias towards features with lots of missing data: Missing values in the target variable are left out of impurity calculations. This provided generally good results at a fraction of the computational costs of imputing data. Optionally, feature.ImputeMissing or featurematrixImputeMissing can be called before forest growth to impute missing values to the feature mean/mode which Brieman [2] suggests as a fast method for imputing values. This forest could also be analyzed for proximity (using leafcount or tree.GetLeaves) to do the more accurate proximity weighted imputation Brieman describes. Experimental support is provided for 3 way splitting which splits missing cases onto a third branch. [2] This has so far yielded mixed results in testing. At some point in the future support may be added for local imputing of missing values during tree growth as described in [3] [1] http://www.stat.berkeley.edu/~breiman/RandomForests/cc_home.htm#missing1 [2] https://code.google.com/p/rf-ace/ [3] http://projecteuclid.org/DPubS?verb=Display&version=1.0&service=UI&handle=euclid.aoas/1223908043&page=record In CloudForest data is stored using the FeatureMatrix struct which contains Features. The Feature struct implements storage and methods for both categorical and numerical data and calculations of impurity etc and the search for the best split. The Target interface abstracts the methods of Feature that are needed for a feature to be predictable. This allows for the implementation of alternative types of regression and classification. Trees are built from Nodes and Splitters and stored within a Forest. Tree has a Grow implements Brieman and Cutler's method (see extract above) for growing a tree. A GrowForest method is also provided that implements the rest of the method including sampling cases but it may be faster to grow the forest to disk as in the growforest utility. Prediction and Voting is done using Tree.Vote and CatBallotBox and NumBallotBox which implement the VoteTallyer interface.

The sockdrawer command is an analysis and visualization tool to help you reorganize a complex Go package into several simpler ones. sockdrawer operates on three kinds of graphs at different levels of abstraction. The lowest level is the NODE GRAPH. A node is a package-level declaration of a named entity (func, var, const or type). An entire constant declaration is treated as a single node, even if it contains multiple "specs" each defining multiple names, since constants so grouped are typically closely related; an important special case is an enumerated set data type. Also, we treat each "spec" of a var or type declaration as a single node. Each reference to a package-level entity E forms an edge in the node graph, from the node in which it appears to the node E. For example: Each method declaration depends on its receiver named type; in addition we add an edge from each receiver type to its methods: to ensure that a type and its methods stay together. The node graph is highly cyclic, and obviously all nodes in a cycle must belong to the same package for the package import graph to remain acyclic. So, we compute the second graph, the SCNODE GRAPH. In essence, the scnode graph is the graph of strongly connected components (SCCs) of the (ordinary) node graph. By construction, the scnode graph is acyclic. We optionally perform an optimization at this point, which is to fuse single-predecessor scnodes with their sole predecessor, as this tends to reduce clutter in big graphs. This means that the scnodes are no longer true SCCs; however, the scnode graph remains acyclic. We define a valid PARTITION P of the scnode graph as a mapping from scnodes to CLUSTERS such that the projection of the scnode graph using mapping P is an acyclic graph. This third graph is the CLUSTER GRAPH. Every partition represents a valid refactoring of the original package into hypothetical subpackages, each cluster being a subpackage. Two partitions define the extreme ends of a spectrum: the MINIMAL partition maps every scnode to a single cluster; it represents the status quo, a monolithic package. The MAXIMAL partition maps each scnode to a unique cluster; this breaks the package up into an impractically large number of small fragments. The ideal partition lies somewhere in between. The --clusters=<file> argument specifies a CLUSTERS FILE that constrains the partition algorithm. The file consists of a number of stanzas, each assigning an import path to a cluster ("mypkg/internal/util") and assigning a set of initial nodes ({x, y, z}) to it: Order of stanzas is important: clusters must be be declared bottom to top. After each stanza, all nodes transitively reachable (via the node graph) from that cluster are assigned to that cluster, if they have not yet been assigned to some other cluster. Thus we need only mention the root nodes of the cluster, not all its internal nodes. A warning is reported if a node mentioned in a stanza already belongs to a previously defined cluster. There is an implicit cluster, "residue", that holds all remaining nodes after the clusters defined by the file have been processed. Initially, when the clusters file is empty, the residue cluster contains the entire package. (It is logically at the top.) The task for the user is to iteratively define new clusters until the residue becomes empty. When sockdrawer is run, it analyzes the source package, builds the node graph and the scgraph, loads the clusters file, computes the clusters for every node, and then emits SVG renderings of the three levels of graphs, with nodes colors coded as follows: The graphs of all clusters, a DAG, has green nodes; clicking one takes you to the graph over scnodes for that cluster, also a DAG. Each pink node in this graph represents a cyclical bunch of the node graph, collapsed together for ease of viewing. Each blue node here represents a singleton SCC, a single declaration; singular SCCs are replaced by their sole element for simplicity. Clicking a pink (plural) scnode shows the cyclical portion of the node graph that it represents. (If the fusion optimization was enabled, it may not be fully cyclic.) All of its nodes are blue. Clicking a blue node shows the definition of that node in godoc. (The godoc server's base URL is specified by the --godoc flag.) Initially, all nodes belong to the "residue" cluster. (GraphViz graph rendering can be slow for the first several iterations. A large monitor is essential.) The sockdrawer user's task when decomposing a package into clusters is to identify the lowest-hanging fruit (so to speak) in the residue cluster---a coherent group of related scnodes at the bottom of the graph---and to "snip off" a bunch at the "stem" by appending a new stanza to the clusters file and listing the roots of that bunch in the stanza, and then to re-run the tool. Nodes may be added to an existing stanza if appropriate, but if they are added to a cluster that is "too low", this may create conflicts; keep an eye out for warnings. This process continues iteratively until the residue has become empty and the sets of clusters are satisfactory. The tool prints the assignments of nodes to clusters: the "shopping list" for the refactoring work. Clusters should be split off into subpackages in dependency order, lowest first. The analysis chooses a single configuration, such as linux/amd64. Declarations for other configurations (e.g. windows/arm) will be absent from the node graph. There may be some excessively large SCCs in the node graph that reflect a circularity in the design. For the purposes of analysis, you can break them arbitrarily by commenting out some code, though more thought will be required for a principled fix (e.g. dependency injection).

Package pbc provides structures for building pairing-based cryptosystems. It is a wrapper around the Pairing-Based Cryptography (PBC) Library authored by Ben Lynn (https://crypto.stanford.edu/pbc/). This wrapper provides access to all PBC functions. It supports generation of various types of elliptic curves and pairings, element initialization, I/O, and arithmetic. These features can be used to quickly build pairing-based or conventional cryptosystems. The PBC library is designed to be extremely fast. Internally, it uses GMP for arbitrary-precision arithmetic. It also includes a wide variety of optimizations that make pairing-based cryptography highly efficient. To improve performance, PBC does not perform type checking to ensure that operations actually make sense. The Go wrapper provides the ability to add compatibility checks to most operations, or to use unchecked elements to maximize performance. Since this library provides low-level access to pairing primitives, it is very easy to accidentally construct insecure systems. This library is intended to be used by cryptographers or to implement well-analyzed cryptosystems. Cryptographic pairings are defined over three mathematical groups: G1, G2, and GT, where each group is typically of the same order r. Additionally, a bilinear map e maps a pair of elements — one from G1 and another from G2 — to an element in GT. This map e has the following additional property: If G1 == G2, then a pairing is said to be symmetric. Otherwise, it is asymmetric. Pairings can be used to construct a variety of efficient cryptosystems. The PBC library currently supports 5 different types of pairings, each with configurable parameters. These types are designated alphabetically, roughly in chronological order of introduction. Type A, D, E, F, and G pairings are implemented in the library. Each type has different time and space requirements. For more information about the types, see the documentation for the corresponding generator calls, or the PBC manual page at https://crypto.stanford.edu/pbc/manual/ch05s01.html. This package must be compiled using cgo. It also requires the installation of GMP and PBC. During the build process, this package will attempt to include <gmp.h> and <pbc/pbc.h>, and then dynamically link to GMP and PBC. Most systems include a package for GMP. To install GMP in Debian / Ubuntu: For an RPM installation with YUM: For installation with Fink (http://www.finkproject.org/) on Mac OS X: For more information or to compile from source, visit https://gmplib.org/ To install the PBC library, download the appropriate files for your system from https://crypto.stanford.edu/pbc/download.html. PBC has three dependencies: the gcc compiler, flex (http://flex.sourceforge.net/), and bison (https://www.gnu.org/software/bison/). See the respective sites for installation instructions. Most distributions include packages for these libraries. For example, in Debian / Ubuntu: The PBC source can be compiled and installed using the usual GNU Build System: After installing, you may need to rebuild the search path for libraries: It is possible to install the package on Windows through the use of MinGW and MSYS. MSYS is required for installing PBC, while GMP can be installed through a package. Based on your MinGW installation, you may need to add "-I/usr/local/include" to CPPFLAGS and "-L/usr/local/lib" to LDFLAGS when building PBC. Likewise, you may need to add these options to CGO_CPPFLAGS and CGO_LDFLAGS when installing this package. This package is free software: you can redistribute it and/or modify it under the terms of the GNU Lesser General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version. For additional details, see the COPYING and COPYING.LESSER files. This example generates a pairing and some random group elements, then applies the pairing operation. This example computes and verifies a Boneh-Lynn-Shacham signature in a simulated conversation between Alice and Bob.

Package hdrhist provides high dynamic range (HDR) histograms. HDR histograms can be used to accurately record and analyze distributions with very large ranges of data. This HDR histogram implementation is meant to have minimal memory usage and support recording values quickly. With a precision of 3 significant digits and a range of 1000-100 billion (e.g. 1 μs to 100 s), an HDR histogram consumes about 156 KB. Benchmarks show that recording a value in a histogram should take about 12 ns depending on the hardware. A typical usecase for HDR Histograms would be recording latency values in client or server software. This package is a Go port of Gil Tene's HdrHistogram Java package (http://hdrhistogram.github.io/HdrHistogram/). Package hdrhist aims to interoperate with HdrHistogram to the fullest extent possible. Unless otherwise noted, none of the types in the package are safe for concurrent use.

Package pbc provides structures for building pairing-based cryptosystems. It is a wrapper around the Pairing-Based Cryptography (PBC) Library authored by Ben Lynn (https://crypto.stanford.edu/pbc/). This wrapper provides access to all PBC functions. It supports generation of various types of elliptic curves and pairings, element initialization, I/O, and arithmetic. These features can be used to quickly build pairing-based or conventional cryptosystems. The PBC library is designed to be extremely fast. Internally, it uses GMP for arbitrary-precision arithmetic. It also includes a wide variety of optimizations that make pairing-based cryptography highly efficient. To improve performance, PBC does not perform type checking to ensure that operations actually make sense. The Go wrapper provides the ability to add compatibility checks to most operations, or to use unchecked elements to maximize performance. Since this library provides low-level access to pairing primitives, it is very easy to accidentally construct insecure systems. This library is intended to be used by cryptographers or to implement well-analyzed cryptosystems. Cryptographic pairings are defined over three mathematical groups: G1, G2, and GT, where each group is typically of the same order r. Additionally, a bilinear map e maps a pair of elements — one from G1 and another from G2 — to an element in GT. This map e has the following additional property: If G1 == G2, then a pairing is said to be symmetric. Otherwise, it is asymmetric. Pairings can be used to construct a variety of efficient cryptosystems. The PBC library currently supports 5 different types of pairings, each with configurable parameters. These types are designated alphabetically, roughly in chronological order of introduction. Type A, D, E, F, and G pairings are implemented in the library. Each type has different time and space requirements. For more information about the types, see the documentation for the corresponding generator calls, or the PBC manual page at https://crypto.stanford.edu/pbc/manual/ch05s01.html. This package must be compiled using cgo. It also requires the installation of GMP and PBC. During the build process, this package will attempt to include <gmp.h> and <pbc/pbc.h>, and then dynamically link to GMP and PBC. Most systems include a package for GMP. To install GMP in Debian / Ubuntu: For an RPM installation with YUM: For installation with Fink (http://www.finkproject.org/) on Mac OS X: For more information or to compile from source, visit https://gmplib.org/ To install the PBC library, download the appropriate files for your system from https://crypto.stanford.edu/pbc/download.html. PBC has three dependencies: the gcc compiler, flex (http://flex.sourceforge.net/), and bison (https://www.gnu.org/software/bison/). See the respective sites for installation instructions. Most distributions include packages for these libraries. For example, in Debian / Ubuntu: The PBC source can be compiled and installed using the usual GNU Build System: After installing, you may need to rebuild the search path for libraries: It is possible to install the package on Windows through the use of MinGW and MSYS. MSYS is required for installing PBC, while GMP can be installed through a package. Based on your MinGW installation, you may need to add "-I/usr/local/include" to CPPFLAGS and "-L/usr/local/lib" to LDFLAGS when building PBC. Likewise, you may need to add these options to CGO_CPPFLAGS and CGO_LDFLAGS when installing this package. This package is free software: you can redistribute it and/or modify it under the terms of the GNU Lesser General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version. For additional details, see the COPYING and COPYING.LESSER files. This example generates a pairing and some random group elements, then applies the pairing operation. This example computes and verifies a Boneh-Lynn-Shacham signature in a simulated conversation between Alice and Bob.

Package kind provides utilities to analyze and represent the type details of Go values. The Kind struct represents the type details of an instance, including simple types like int, uint, string, bool, etc., as well as complex types like slices, maps, structs, arrays, interfaces, and pointers. The package exposes the Of function that takes a value as input and returns a Kind instance representing the type of that value. The Kind struct contains various boolean flags and methods to query the type details. It also provides a function, deepEqualKind, to compare two Kind instances for equality, allowing users to check if two values have the same type and value representation. Example usage: The Kind struct has methods like IsComplex, IsSigned, IsUnsigned, IsNil, IsPointer, etc., to query various attributes of the type. It also provides As<Type>() methods to extract the underlying value as a specific Go type (e.g., AsBool, AsString, AsInt, AsFloat32, etc.). Additionally, Kind provides MapKeyKind and MapValueKind methods to get the Kind instances of the key and value types of a map, respectively. The package is useful for situations where type introspection or reflection is needed to understand the nature of Go values, especially in cases where nested or complex types are involved. Note: The package assumes that the value passed to the Of function is a valid Go value, and it does not handle all possible types. For instance, certain edge cases like unsafe pointers, function pointers, and unexported fields in structs might not be fully supported. Users should exercise caution and test thoroughly in their specific use cases. This package is written in pure Go and does not have any external dependencies.