markdown_link_checker_sc 0.0.116 → 0.0.118

package/index.js

@@ -48,7 +48,7 @@ program
   )
   .option(
     "-l, --log <types...>",
-    "Export logs for debugging. Types: allerrors, filterederrors, allresults etc."
+    "Types of console logs to display logs for debugging. Types: functions, todo etc."
   )
   .option(
     "-f, --files <path>",
@@ -64,6 +64,11 @@ program
     "-u, --site_url [value]",
     "Site base url in form dev.example.com (used to catch absolute urls to local files)"
   )
+  .option(
+    "-o, --logtofile [value]",
+    "Output logs to file",
+    true
+  )
 
   .parse(process.argv);
 
@@ -78,39 +83,30 @@ sharedData.allHTMLFiles = new Set([]);
 sharedData.allImageFiles = new Set([]);
 sharedData.allOtherFiles = new Set([]);
 
-
 const markdownDirectory = path.join(sharedData.options.root, sharedData.options.directory);
-if (sharedData.options.log == "fast") {
-  console.log(`MARKDOWN DIR ${markdownDirectory}`);
-}
-
-async () => {
-  // Load JSON file containing file paths and reassign as array to the JSON path
-  sharedData.options.files
-    ? (sharedData.options.files = await loadJSONFileToReportOn(sharedData.options.files))
-    : (sharedData.options.files = []);
-  if (sharedData.options.log == "quick") {
-    for (const file of sharedData.options.files) {
-      console.log(file);
-    }
-  }
-};
 
 // Function for loading JSON file that contains files to report on
 async function loadJSONFileToReportOn(filePath) {
   sharedData.options.log.includes("functions")
     ? console.log(`Function: loadJSONFileToReportOn(): filePath: ${filePath}`)
     : null;
+    sharedData.options.log.includes("quick")
+    ? console.log(`Function: loadJSONFileToReportOn(): filePath: ${filePath}`)
+    : null;
   try {
     const fileContent = await fs.promises.readFile(filePath, "utf8");
     let filesArray = JSON.parse(fileContent);
     // Array relative to root, so update to have full path
     filesArray = filesArray.map((str) => path.join(sharedData.options.root, str));
+    
+    sharedData.options.log.includes("quick")
+    ? console.log(`quick:filesArray: ${filesArray}`)
+    : null;
 
-    //console.log(filesArray);
     return filesArray;
   } catch (error) {
     console.error(`Error reading file: ${error.message}`);
+    console.log(`Error reading file: ${error.message}`);
     process.exit(1);
   }
 }
@@ -193,12 +189,12 @@ function filterErrors(errors) {
   // Filter results on specified file names (if any specified)
   //console.log(`Number pages to filter: ${sharedData.options.files.length}`);
   if (sharedData.options.files.length > 0) {
+	  //console.log(`USharedFileslength: ${sharedData.options.files.length}`);
     filteredErrors = errors.filter((error) => {
       //console.log(`UError: ${error}`);
       //console.log(JSON.stringify(error, null, 2));
-      //console.log(`UError page: ${error.page}`);
-      const filterResult = sharedData.options.files.includes(error.page);
-      //console.log(`filterResult: ${filterResult}`);
+      //console.log(`UError file: ${error.file}`);
+      const filterResult = sharedData.options.files.includes(error.file);
       return filterResult;
     });
   }
@@ -210,6 +206,9 @@ function filterErrors(errors) {
 
 //main function, after options et have been set up.
 (async () => {
+
+  sharedData.options.files ? (sharedData.options.files = await loadJSONFileToReportOn(sharedData.options.files)) : (sharedData.options.files = []);
+
   // process  containing markdown, return results which includes links, headings, id anchors
   const results = await processDirectory(markdownDirectory);
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "markdown_link_checker_sc",
-  "version": "0.0.116",
+  "version": "0.0.118",
   "description": "Markdown Link Checker",
   "main": "index.js",
   "scripts": {

package/src/helpers.js

@@ -1,9 +1,26 @@
 import fs from "fs/promises";
 import path from "path";
+import { sharedData } from "./shared_data.js";
+
+// Call at start of function to ease logging to console.
+function logFunction(name, ...args) {
+  sharedData.options.log.includes("functions")
+    ? console.log(`${name}: ${args.join(", ")}`)
+    : null;
+}
 
 // Log data to specified file path, replacing file.
 async function logToFile(filePath, dataString) {
+  sharedData.options.log.includes("functions")
+    ? console.log(`Function: logToFile(${filePath}, dataString))`)
+    : null;
+  if (!sharedData.options.logtofile) {
+    return;
+  }
+
   try {
+    await fs.mkdir(path.dirname(filePath), { recursive: true });
+
     await fs.writeFile(filePath, dataString);
     //console.log("Data written to file");
   } catch (err) {
@@ -38,4 +55,4 @@ function isHTML(file) {
   return fileExtension === ".html" || fileExtension === ".htm" ? true : false;
 }
 
-export { logToFile, isImage, isMarkdown, isHTML };
+export { logToFile, logFunction, isImage, isMarkdown, isHTML };

package/src/links.js

@@ -1,6 +1,7 @@
 import path from "path";
 import { isImage, isMarkdown, isHTML } from "./helpers.js";
 import { sharedData } from "./shared_data.js";
+import { logFunction } from "./helpers.js";
 
 class Link {
   address = "";
@@ -33,9 +34,7 @@ class Link {
   }
 
   constructor({ page, url, type, text, title }) {
-    sharedData.options.log.includes("functions")
-    ? console.log("Link:constructor")
-    : null;
+    logFunction("Link:constructor");
 
     if (page) {
       this.page = page;
@@ -110,9 +109,8 @@ class Link {
   // This is only used if the type is not specified as an argument.
   // Uses file extension etc, so should be run after SplitUrl() which finds the address
   findType() {
-    sharedData.options.log.includes("functions")
-    ? console.log("Link:findType()")
-    : null;
+    logFunction("Link:findType()");
+
     let linkType = "unHandledLinkType";
 
     this.isImage = this.address && isImage(this.address) ? true : false; //only if address is true.
@@ -182,9 +180,7 @@ class Link {
 
   //get absolute path to link, if this is a relative URL link.
   getAbsolutePath() {
-    sharedData.options.log.includes("functions")
-    ? console.log(`Link:getAbsolutePath: page: ${this.page}, address: ${this.address} `)
-    : null;
+    logFunction("Link:getAbsolutePath", `this.page: ${this.page}`, `this.address: ${this.address}`);
     if (!this.isRelative) throw new Error("Link:getAbsolutePath() called on non-relative path");
     return path.resolve(path.dirname(this.page), this.address);
   }

package/src/output_errors.js

@@ -1,14 +1,13 @@
 //import { /*LinkError,*/ CurrentFileMissingAnchorError, LinkedFileMissingAnchorError, LinkedInternalPageMissingError, InternalLinkToHTMLError, UrlToLocalSiteError} from "./errors.js"
 
 import { sharedData } from "./shared_data.js";
-
+import { logFunction } from "./helpers.js";
 
 //Function that generates console and/or log output from an array of error objects.
-// - `results` is an array of error objects. These will have a `type` and a `page`. They may also have other values, depending on type of error - such as linkurl
+// - `results` is an array of error objects.
+//  These will have a `type` and a `page`. They may also have other values, depending on type of error - such as linkurl
 function outputErrors(results) {
-  sharedData.options.log.includes("functions")
-    ? console.log("Function: outputErrors()")
-    : null;
+  logFunction("Function: outputErrors()");
 
   //Sort results by page and type.
   // Perhaps next step is to create only get info for particular pages.

package/src/process_image_orphans.js

@@ -3,6 +3,7 @@ import fs from "fs";
 import path from "path";
 import { sharedData } from "./shared_data.js";
 import { OrphanedImageError } from "./errors.js";
+import { logFunction } from "./helpers.js";
 
 function isImage(file) {
   const imageExtensions = [".jpg", ".jpeg", ".png", ".svg", ".gif", ".webm"];
@@ -14,9 +15,7 @@ var otherFileTypes = []; // Just used for logging in function below.
 
 // Gets all image files in a directory.
 async function getAllImageFilesInDirectory(dir) {
-  sharedData.options.log.includes("functions")
-    ? console.log(`Function: getAllImageFilesInDirectory(${dir})`)
-    : null;
+  logFunction(`Function: getAllImageFilesInDirectory(${dir})`)
 
   // TODO put this all in a try catch and return a better error.
   // Or perhaps put around parent.
@@ -43,9 +42,8 @@ async function getAllImageFilesInDirectory(dir) {
 
 // Checks if any images in the options.directory
 async function checkImageOrphansGlobal(results) {
-  sharedData.options.log.includes("functions")
-    ? console.log("Function: checkImageOrphansGlobal")
-    : null;
+  logFunction(`Function: checkImageOrphansGlobal()`)
+  
   const errors = [];
   let allImagesFound = [];
 

package/src/process_internal_url_links.js

@@ -1,8 +1,10 @@
 import path from "path";
 import { UrlToLocalSiteError} from "./errors.js"
+import { logFunction } from "./helpers.js";
 
 // An array of errors given a results object that contains our array of objects containing urls that link to our current site.
 function processUrlsToLocalSource(results) {
+  logFunction(`Function: processUrlsToLocalSource()`);
   const errors = [];
   results.forEach((page, index, array) => {
     //console.log(`PAGE: ${page}`);

package/src/process_local_image_links.js

@@ -2,11 +2,12 @@ import path from "path";
 import fs from "fs";
 import { sharedData } from "./shared_data.js";
 import { LocalImageNotFoundError } from "./errors.js";
+import { logFunction } from "./helpers.js";
 
 // Checks if every image in every markdown page (results.page.relativeImageLinks) is present on the file system.
 // - results is the array of information coming out of markdown parsing.
 async function checkLocalImageLinks(results) {
-  sharedData.options.log.includes("functions") ? console.log(`Function: checkLocalImageLinks()`) : null;
+  logFunction(`Function: checkLocalImageLinks()`);
   const errors = [];
   const promises = [];
 
@@ -20,25 +21,13 @@ async function checkLocalImageLinks(results) {
       //console.log(`link.linkUrlt: ${link.url}`);
       //console.log(`dirname: ${path.dirname(page.page_file)}`);
 
-      const fullImagePath = path.join(
-        path.dirname(page.page_file),
-        link.url
-      );
+      const fullImagePath = path.join(path.dirname(page.page_file), link.url);
       //console.log(`fullImagePath: ${fullImagePath}`);
       const promise = new Promise((resolve) => {
         fs.access(fullImagePath, fs.constants.F_OK, (err) => {
           if (err) {
             //console.log("Error");
-			      const error = new LocalImageNotFoundError({link: link});
-            /*
-            const error = {
-              type: "LocalImageNotFound",
-              page: `${page.page_file}`,
-              linkUrl: `${link.url}`,
-              linkText: `${link.text}`,
-              linkFullPath: `${fullImagePath}`,
-            };
-            */
+            const error = new LocalImageNotFoundError({ link: link });
             errors.push(error);
             resolve(false);
           } else {

package/src/process_markdown.js

@@ -1,12 +1,12 @@
 import { Link } from "./links.js";
 import { sharedData } from "./shared_data.js";
+import { logFunction } from "./helpers.js";
 
 // Returns slug for a string (markdown heading) using Vuepress algorithm.
 // Algorithm from chatgpt - needs testing.
 const processMarkdown = (contents, page) => {
-  sharedData.options.log.includes("functions")
-    ? console.log(`Function: processMarkdown(): page: ${page}`)
-    : null;
+  logFunction(`Function: processMarkdown(): page: ${page}`);
+
   const headings = [];
   //const anchors = [];
   const htmlAnchors = []; //{};
@@ -103,10 +103,10 @@ const processLineMarkdownLinks = (
   unHandledLinkTypes,
   page
 ) => {
-  sharedData.options.log.includes("functions")
-    ? console.log(`Function: processLineMarkdownLinks(): page: ${page}`)
-    : null;
+  logFunction(`Function: processMarkdown(): page: ${page}`);
+
   //const regex = /(?<prefix>[!@]?)\[(?<text>[^\]]+)\]\((?<url>\S+?)(?:\s+"(?<title>[^"]+)")?\)/g;
+  // Match to Markdown link OR image
   const regex =
     /(?<prefix>[!@]?)\[(?<text>[^\]]*)\]\((?<url>\S+?)(?:\s+"(?<title>[^"]+)")?\)/g;
   const matches = line.matchAll(regex);
@@ -310,12 +310,16 @@ const processLineMarkdownLinks = (
   //Might further parse this to catch img in anchor.
 
   //Match for html img - append to the lists
-  const regexHTMLImgTotal = /<img\s+(?<attributes>.*?)\/>/gi;
+  const regexHTMLImgTotal = /<img\s+(?<attributes>.*?)>/gi;
+  //const regexHTMLImgTotal = /<img\s+(?<attributes>.*?)\/>/gi;
+
   const regex_htmlattr_src =
     /src\s*[=]\s*(?<quote>['"])(?<src>.*?)(?<!\\)\k<quote>/i;
+
+
   for (const match of line.matchAll(regexHTMLImgTotal)) {
+    //console.log(`XXXXXregexHTMLImgTotals: ${match}`)
     const attributes = match.groups.attributes;
-    //console.log(`XXXXXImageattributes_s: ${attributes}`)
     const linkText = "";
     let linkTitle = "";
     let linkUrl = "";

package/src/process_orphans.js

@@ -2,14 +2,13 @@ import { logToFile } from "./helpers.js";
 import path from "path";
 import { sharedData } from "./shared_data.js";
 import { PageNotInTOCError, PageNotLinkedInternallyError } from "./errors.js";
-
+import { logFunction } from "./helpers.js";
 
 // Gets page with most links. Supposed to be used on the allResults object that is an array of objects about each page.
 // Will use to get the summary.
 function getPageWithMostLinks(pages) {
-  if (sharedData.options.log.includes("functions")) {
-    console.log("Function: getPageWithMostLinks");
-  }
+  logFunction(`Function: getPageWithMostLinks`);
+
   return pages.reduce(
     (maxLinksPage, currentPage) => {
       if (

package/src/process_relative_links.js

@@ -7,12 +7,11 @@ import {
   UrlToLocalSiteError,
 } from "./errors.js";
 import { sharedData } from "./shared_data.js";
+import { logFunction } from "./helpers.js";
 
 // An array of errors given a results object that contains our array of objects containing relativeLinks (and other information).
 function processRelativeLinks(results) {
-  sharedData.options.log.includes("functions")
-    ? console.log("Function: processRelativeLinks")
-    : null;
+  logFunction(`Function: processRelativeLinks()`);
   const errors = [];
 
   //console.log(sharedData);

package/tests/links/html/img/tests1.md

@@ -0,0 +1,12 @@
+# Test
+
+Run like: `node .\index.js -d tests/links/html/img/`
+
+Confirm links with and without end backslash are caught.
+Also with/without attributes
+
+<img src="../../assets/airframes/multicopter/x500_holybro_pixhawk4/img_with_no_end_backslash_oneofthree.png" width="400" title="Hanger gaskets">
+
+<img src="../../assets/airframes/multicopter/x500_holybro_pixhawk4/img_with_backslash_twoofthree.png" width="400" title="Hanger gaskets" />
+
+<img src="img_with_backslash_no_attributes_threeofthree.png" />