word-extractor - npm Package Compare versions

Comparing version 1.0.2 to 1.0.3

CHANGELOG.md

 # Change log
 
+### 1.0.3 / 17th June2021
+
+ * Fixes issues with long attribute values (> 65k) in OO XML. See #37
+ * Propogate errors from XML failures into promise rejections. See #38
+ * Changed the XML parser dependency for maintenance and fixes. See #39
+ 
 ### 1.0.2 / 28th May 2021
@@ -4,0 +10,0 @@

lib/open-office-extractor.js

@@ -20,3 +20,3 @@
 const path = require('path');
-const SAX = require("sax");
+const SAXES = require("saxes");
 const yauzl = require('yauzl');
@@ -100,12 +100,10 @@
           zipfile.readEntry();
-          zipfile.on("error", function(error) {
-            reject(error);
-          });
+          zipfile.on("error", reject);
           zipfile.on("entry", (entry) => {
             if ('[Content_Types].xml' === entry.fileName || this.shouldProcess(entry.fileName)) {
-              //console.log("entry", entry.fileName);
               return this.handleEntry(zipfile, entry)
                 .then(() => {
                   zipfile.readEntry();
-                });
+                })
+                .catch((e) => reject(e));
             }
@@ -115,6 +113,3 @@
           });
-          zipfile.on("end", () => {
-            //console.log(this._document);
-            resolve(this._document);
-          });  
+          zipfile.on("end", () => resolve(this._document));
         });
@@ -134,121 +129,134 @@ })
 
+  handleOpenTag(node) {
+    if (node.name === 'Override') {
+      const actionFunction = this._streamTypes[node.attributes['ContentType']];
+      if (actionFunction) {
+        const partName = node.attributes['PartName'].replace(/^[/]+/, '');
+        const action = {action: actionFunction, type: node.attributes['ContentType']};
+        this._actions[partName] = action;
+      }
+    } else if (node.name === 'Default') {
+      const extension = node.attributes['Extension'];
+      const contentType = node.attributes['ContentType'];
+      this._defaults[extension] = contentType;
+    } else if (node.name === 'Relationship') {
+      // console.log(this._source, node);
+      this._relationships[node.attributes['Id']] = {
+        type: node.attributes['Type'],
+        target: node.attributes['Target'],
+      };
+    } else if (node.name === 'w:document' || 
+               node.name === 'w:footnotes' || 
+               node.name === 'w:endnotes' || 
+               node.name === 'w:comments') {
+      this._context = ['content', 'body'];
+      this._pieces = [];
+    } else if (node.name === 'w:hdr' ||
+               node.name === 'w:ftr') {
+      this._context = ['content', 'header'];
+      this._pieces = [];
+    } else if (node.name === 'w:endnote' || node.name === 'w:footnote') {
+      const type = (node.attributes['w:type'] || this._context[0]);
+      this._context.unshift(type);
+    } else if (node.name === 'w:tab' && this._context[0] === 'content') {
+      this._pieces.push("\t");
+    } else if (node.name === 'w:br' && this._context[0] === 'content') {
+      if ((node.attributes['w:type'] || '') === 'page') {
+        this._pieces.push("\n");
+      } else {
+        this._pieces.push("\n");
+      }
+    } else if (node.name === 'w:del' || node.name === 'w:instrText') {
+      this._context.unshift('deleted');
+    } else if (node.name === 'w:tabs') {
+      this._context.unshift('tabs');
+    } else if (node.name === 'w:tc') {
+      this._context.unshift('cell');
+    } else if (node.name === 'w:drawing') {
+      this._context.unshift('drawing');
+    } else if (node.name === 'w:txbxContent') {
+      this._context.unshift(this._pieces);
+      this._context.unshift('textbox');
+      this._pieces = [];
+    }
+  }
+
+  handleCloseTag(node) {
+    if (node.name === 'w:document') {
+      this._context = null;
+      this._document._body = this._pieces.join("");
+    } else if (node.name === 'w:footnote' || node.name === 'w:endnote') {
+      this._context.shift();
+    } else if (node.name === 'w:footnotes') {
+      this._context = null;
+      this._document._footnotes = this._pieces.join("");
+    } else if (node.name === 'w:endnotes') {
+      this._context = null;
+      this._document._endnotes = this._pieces.join("");
+    } else if (node.name === 'w:comments') {
+      this._context = null;
+      this._document._annotations = this._pieces.join("");
+    } else if (node.name === 'w:hdr') {
+      this._context = null;
+      this._document._headers = this._document._headers + this._pieces.join("");
+    } else if (node.name === 'w:ftr') {
+      this._context = null;
+      this._document._footers = this._document._footers + this._pieces.join("");
+    } else if (node.name === 'w:p') {
+      if (this._context[0] === 'content' || this._context[0] === 'cell' || this._context[0] === 'textbox') {
+        this._pieces.push("\n");
+      }
+    } else if (node.name === 'w:del' || node.name === 'w:instrText') {
+      this._context.shift();
+    } else if (node.name === 'w:tabs') {
+      this._context.shift();
+    } else if (node.name === 'w:tc') {
+      this._pieces.pop();
+      this._pieces.push("\t");
+      this._context.shift();
+    } else if (node.name === 'w:tr') {
+      this._pieces.push("\n");
+    } else if (node.name === 'w:drawing') {
+      this._context.shift();
+    } else if (node.name === 'w:txbxContent') {
+      const textBox = this._pieces.join("");
+      const context = this._context.shift();
+      if (context !== 'textbox') {
+        throw new Error("Invalid textbox context");
+      }
+      this._pieces = this._context.shift();
+      
+      // If in drawing context, discard
+      if (this._context[0] === 'drawing')
+        return;
+
+      if (textBox.length == 0)
+        return;
+      
+      const inHeader = this._context.includes('header');
+      const documentField = (inHeader) ? '_headerTextboxes' : '_textboxes';
+      if (this._document[documentField]) {
+        this._document[documentField] = this._document[documentField] + "\n" + textBox;
+      } else {
+        this._document[documentField] = textBox;
+      }
+    }
+  }
+
   createXmlParser() {
-    const strict = true;
-    const parser = SAX.createStream(strict); 
+    const parser = new SAXES.SaxesParser(); 
 
     parser.on("opentag", (node) => {
-      if (node.name === 'Override') {
-        const actionFunction = this._streamTypes[node.attributes['ContentType']];
-        if (actionFunction) {
-          const partName = node.attributes['PartName'].replace(/^[/]+/, '');
-          const action = {action: actionFunction, type: node.attributes['ContentType']};
-          this._actions[partName] = action;
-        }
-      } else if (node.name === 'Default') {
-        const extension = node.attributes['Extension'];
-        const contentType = node.attributes['ContentType'];
-        this._defaults[extension] = contentType;
-      } else if (node.name === 'Relationship') {
-        // console.log(this._source, node);
-        this._relationships[node.attributes['Id']] = {
-          type: node.attributes['Type'],
-          target: node.attributes['Target'],
-        };
-      } else if (node.name === 'w:document' || 
-                 node.name === 'w:footnotes' || 
-                 node.name === 'w:endnotes' || 
-                 node.name === 'w:comments') {
-        this._context = ['content', 'body'];
-        this._pieces = [];
-      } else if (node.name === 'w:hdr' ||
-                 node.name === 'w:ftr') {
-        this._context = ['content', 'header'];
-        this._pieces = [];
-      } else if (node.name === 'w:endnote' || node.name === 'w:footnote') {
-        const type = (node.attributes['w:type'] || this._context[0]);
-        this._context.unshift(type);
-      } else if (node.name === 'w:tab' && this._context[0] === 'content') {
-        this._pieces.push("\t");
-      } else if (node.name === 'w:br' && this._context[0] === 'content') {
-        if ((node.attributes['w:type'] || '') === 'page') {
-          this._pieces.push("\n");
-        } else {
-          this._pieces.push("\n");
-        }
-      } else if (node.name === 'w:del' || node.name === 'w:instrText') {
-        this._context.unshift('deleted');
-      } else if (node.name === 'w:tabs') {
-        this._context.unshift('tabs');
-      } else if (node.name === 'w:tc') {
-        this._context.unshift('cell');
-      } else if (node.name === 'w:drawing') {
-        this._context.unshift('drawing');
-      } else if (node.name === 'w:txbxContent') {
-        this._context.unshift(this._pieces);
-        this._context.unshift('textbox');
-        this._pieces = [];
+      try {
+        this.handleOpenTag(node);
+      } catch (e) {
+        parser.fail(e.message);
       }
-      
     });
 
     parser.on('closetag', (node) => {
-      if (node === 'w:document') {
-        this._context = null;
-        this._document._body = this._pieces.join("");
-      } else if (node === 'w:footnote' || node === 'w:endnote') {
-        this._context.shift();
-      } else if (node === 'w:footnotes') {
-        this._context = null;
-        this._document._footnotes = this._pieces.join("");
-      } else if (node === 'w:endnotes') {
-        this._context = null;
-        this._document._endnotes = this._pieces.join("");
-      } else if (node === 'w:comments') {
-        this._context = null;
-        this._document._annotations = this._pieces.join("");
-      } else if (node === 'w:hdr') {
-        this._context = null;
-        this._document._headers = this._document._headers + this._pieces.join("");
-      } else if (node === 'w:ftr') {
-        this._context = null;
-        this._document._footers = this._document._footers + this._pieces.join("");
-      } else if (node === 'w:p') {
-        if (this._context[0] === 'content' || this._context[0] === 'cell' || this._context[0] === 'textbox') {
-          this._pieces.push("\n");
-        }
-      } else if (node === 'w:del' || node === 'w:instrText') {
-        this._context.shift();
-      } else if (node === 'w:tabs') {
-        this._context.shift();
-      } else if (node === 'w:tc') {
-        this._pieces.pop();
-        this._pieces.push("\t");
-        this._context.shift();
-      } else if (node === 'w:tr') {
-        this._pieces.push("\n");
-      } else if (node === 'w:drawing') {
-        this._context.shift();
-      } else if (node === 'w:txbxContent') {
-        const textBox = this._pieces.join("");
-        const context = this._context.shift();
-        if (context !== 'textbox') {
-          throw new Error("Invalid textbox context");
-        }
-        this._pieces = this._context.shift();
-        
-        // If in drawing context, discard
-        if (this._context[0] === 'drawing')
-          return;
-
-        if (textBox.length == 0)
-          return;
-        
-        const inHeader = this._context.includes('header');
-        const documentField = (inHeader) ? '_headerTextboxes' : '_textboxes';
-        if (this._document[documentField]) {
-          this._document[documentField] = this._document[documentField] + "\n" + textBox;
-        } else {
-          this._document[documentField] = textBox;
-        }
-
+      try {
+        this.handleCloseTag(node);
+      } catch (e) {
+        parser.fail(e.message);
       }
@@ -258,6 +266,10 @@ });
     parser.on('text', (string) => {
-      if (! this._context)
-        return;
-      if (this._context[0] === 'content' || this._context[0] === 'cell' || this._context[0] === 'textbox') {
-        this._pieces.push(string);
+      try {
+        if (! this._context)
+          return;
+        if (this._context[0] === 'content' || this._context[0] === 'cell' || this._context[0] === 'textbox') {
+          this._pieces.push(string);
+        }
+      } catch (e) {
+        parser.fail(e.message);
       }
@@ -278,4 +290,20 @@ });
         const parser = this.createXmlParser();
-        readStream.on("end", resolve);
-        readStream.pipe(parser);
+        parser.on("error", (e) => {
+          readStream.destroy(e);
+          reject(e);
+        });
+        parser.on("end", () => resolve());
+        readStream.on("end", () => parser.close());
+        readStream.on("error", (e) => reject(e));
+        readStream.on("readable", () => {
+          // eslint-disable-next-line no-constant-condition
+          while (true) {
+            const chunk = readStream.read(0x1000);
+            if (chunk === null) {
+              return;
+            }
+      
+            parser.write(chunk);
+          }
+        });
       });  
@@ -282,0 +310,0 @@ });

package.json

 {
   "name": "word-extractor",
-  "version": "1.0.2",
+  "version": "1.0.3",
   "description": "Node.js package to read Word .doc files",
@@ -32,3 +32,3 @@ "main": "lib/word.js",
   "dependencies": {
-    "sax": "^1.2.4",
+    "saxes": "^5.0.1",
     "yauzl": "^2.10.0"
@@ -35,0 +35,0 @@ },

New alerts

License Policy Violation

License

This package is not allowed per your license policy. Review the package's license to ensure compliance.

Found 1 instance in 1 package

Fixed alerts

License Policy Violation

License

This package is not allowed per your license policy. Review the package's license to ensure compliance.

Found 1 instance in 1 package

Improved metrics

Total package byte prevSize

72996

increased by1.08%

Number of package files

19

increased by5.56%

Lines of code

1692

increased by1.62%

Dependency changes

• + Addedsaxes@^5.0.1

• + Addedsaxes@5.0.1(transitive)

• + Addedxmlchars@2.2.0(transitive)

• - Removedsax@^1.2.4

• - Removedsax@1.4.1(transitive)