bonus2: add support for combinig multiple html files into one LaTeX doc

git-svn-id: svn+ssh://atelier.inf.usi.ch/home/bevilj/group-1@210 a672b425-5310-4d7a-af5c-997e18724b81

bonus2/ReadMe.md

@@ -19,7 +19,7 @@ The output jar will be available at this path: `build/libs/bonus2.jar`
 ## Execute
 
 ```
-java -jar bonus2.jar [file1.html] [file2.html] [directory]
+java -jar bonus2.jar [-s, --singlepage] [file1.html] [file2.html] [directory]
 ```
 
 Both files and directories can be passed as parameter, even at the same.
@@ -33,7 +33,10 @@ subdirectories matching the absolute path of the files passed as argument.
 Note that the program will not work with plain html files,
 it requires Jekyll html files instead.
 
-A test file is available for reference under the directory `res`.
+The `-s` or `--singlepage` flag creates one single output LaTeX file
+instead of many. It will be output'ed in the "out" directory.
+
+Test files is available for reference under the directory `res`.
 
 ## Supported tags
 
@@ -62,4 +65,5 @@ A test file is available for reference under the directory `res`.
 * comments are removed
 * jekyll author -> LaTeX author
 * jekyll title -> LaTeX title
+* jekyll highlight -> verbatim
 

bonus2/res/test.html

@@ -8,8 +8,12 @@ title: The Tragedy
 previous-page: pages/cmd/basic/open.html
 next-page: pages/cmd/basic/ls.html
 ---
-<p>
-Did you ever hear the tragedy of <b>Darth Plagueis</b> <i>The Wise</i>?<br>
+
+Did you ever hear the tragedy of <!--
+  ~ Copyright (c) 2018 Bevilacqua Joey.
+  -->
+
+<b>Darth Plagueis</b> <i>The Wise</i>?<br>
 <code>I thought not</code>. It’s not a story the Jedi would tell you.</br>
 It’s a Sith legend. Darth Plagueis was a Dark Lord of the Sith,
 so powerful and so wise he could use the Force to influence the midichlorians to <u>create life…</u><br>
@@ -19,13 +23,6 @@ He became so powerful… the only thing he was afraid of was losing his power, w
 Unfortunately, he taught his apprentice everything he knew, then his apprentice killed him in his sleep.<br>
 <b>Ironic</b>. He could save others from death, but not himself.<br>
 
-<h1>Surprise</h1>
-
-<pre>
-A surprise to be sure,
-    but a welcome one.
-</pre>
-
 <table>
     <tr>
         <td>Take a</td>
@@ -44,9 +41,7 @@ A surprise to be sure,
 
 <h1>Fish</h1>
 
-There's always <a href="{{ site.baseurl }}/bigger.html">bigger</a> fish<br>
-
-[visible confusion]
+There's always <a href="{{ site.baseurl }}/bigger.html">bigger</a> fish
 
 <ul>
     <li><u>Hello there</u></li>
@@ -78,6 +73,4 @@ Dew it.
         <td>that's a</td>
         <td>good trick</td>
     </tr>
-</table>
-
-</p>
+</table>

bonus2/res/test2.html

@@ -0,0 +1,85 @@
+---
+layout: page
+category-title: Knowledge
+category-page: basic
+tags: tragedy know
+author: Sheev Palpatine
+title: The Tragedy
+previous-page: pages/cmd/basic/open.html
+next-page: pages/cmd/basic/ls.html
+---
+
+Did you ever hear the tragedy of <!--
+  ~ Copyright (c) 2018 Bevilacqua Joey.
+  -->
+
+<b>Darth Plagueis</b> <i>The Wise</i>?<br>
+<code>I thought not</code>. It’s not a story the Jedi would tell you.</br>
+It’s a Sith legend. Darth Plagueis was a Dark Lord of the Sith,
+so powerful and so wise he could use the Force to influence the midichlorians to <u>create life…</u><br>
+He had such a knowledge of the dark side that he could even keep the ones he cared about from dying.<br>
+The dark side of the Force is a pathway to many abilities some consider to be unnatural.<br>
+He became so powerful… the only thing he was afraid of was losing his power, which eventually, of course, he did.<br>
+Unfortunately, he taught his apprentice everything he knew, then his apprentice killed him in his sleep.<br>
+<b>Ironic</b>. He could save others from death, but not himself.<br>
+
+<h1>Surprise</h1>
+
+<pre>
+A surprise to be sure,
+    but a welcome one.
+</pre>
+
+<table>
+    <tr>
+        <td>Take a</td>
+        <td><b>seat</b></td>
+    </tr>
+    <tr>
+        <td><a href="https://google.com">This</a> is where</td>
+        <td>The fun begins</td>
+    </tr>
+</table>
+
+
+<h3>Fish</h3>
+
+<h2>Fish</h2>
+
+<h1>Fish</h1>
+
+There's always <a href="{{ site.baseurl }}/bigger.html">bigger</a> fish<br>
+
+[visible confusion]
+
+<ul>
+    <li><u>Hello there</u></li>
+    <li>General Kenobi</li>
+    <li>You're a <b>bold</b> one</li>
+</ul>
+
+{% highlight bash %}
+
+Dew it.
+
+{% endhighlight %}
+
+
+<table>
+    <tr>
+        <th>The</th>
+        <th>negotiations</th>
+    </tr>
+    <tr>
+        <td>where</td>
+        <td>short</td>
+    </tr>
+    <tr>
+        <td>I'll try</td>
+        <td><i>spinning</i></td>
+    </tr>
+    <tr>
+        <td>that's a</td>
+        <td>good trick</td>
+    </tr>
+</table>

bonus2/src/ch/usi/inf/atelier/group1/HtmlParser.kt

@@ -8,29 +8,48 @@ import ch.usi.inf.atelier.group1.jekyll.HtmlToLatexWriter
 import ch.usi.inf.atelier.group1.jekyll.JekyllPage
 import ch.usi.inf.atelier.group1.util.Log
 import ch.usi.inf.atelier.group1.util.extensions.insertJekyllHeader
-import ch.usi.inf.atelier.group1.util.extensions.writeTo
 import java.io.File
+import java.io.FileWriter
+import java.lang.StringBuilder
+import java.text.SimpleDateFormat
+import java.util.*
 
-class HtmlParser(private val input: File) {
-    private val file = JekyllPage(input)
-    private val output = HtmlToLatexWriter(file.content)
+class HtmlParser(private val singlePage: Boolean) {
+    private val content = StringBuilder()
+    private var outName = "${SimpleDateFormat("yyyy-MM-dd_hh:mm").format(Date())}.html"
 
-    fun parse(): String {
+    /**
+     * Parse a jekyll html file to a LaTex Document
+     *
+     * @param path The path of the html file
+     */
+    fun parse(path: File) {
+        val file = JekyllPage(path)
+        val output = HtmlToLatexWriter(file.content, singlePage)
+
+        // Make sure this is a jekyll html file
         if (!file.isValid()) {
-            Log.e(IllegalArgumentException("This file is not valid"))
+            Log.e(IllegalArgumentException("This file is not valid"), false)
+            return
         }
 
-        /*
-        if (file.header["author"] == "Marwan Announ") {
-            throw IllegalStateException("Invalid fuckery. Please don\'t attemp to parse this shit. Kthxbye")
-        }
-        */
-
         output.run {
-            start()
-            insertJekyllHeader(file)
-            beginDocument()
+            if (singlePage) {
+                // In singlePage mode, insert the beginning of the document only once
+                if (content.isEmpty()) {
+                    start()
+                    addSinglePageInfo("Documentation", "Group 1")
+                    beginDocument()
+                }
 
+                insertJekyllHeader(file)
+            } else {
+                start()
+                insertJekyllHeader(file)
+                beginDocument()
+            }
+
+            // Convert html elements
             changeBold()
             changeBr()
             changeCode()
@@ -46,14 +65,60 @@ class HtmlParser(private val input: File) {
             changeTable()
             changeUnderline()
 
+            // Strip html comments
             stripComments()
 
-            endDocument()
+            // Store the converted document
+            commit()
 
-            writeTo(input.absolutePath.replace(".html", ".tex"))
+            if (!singlePage) {
+                // End the document if singlePage is not enabled
+                endDocument()
+            }
         }
 
-        return output.toString()
+        Log.i("Parsed: $path")
+
+        content.append(output.toString())
+
+        // Set the file name basing on the original html file name if not running in singlePage mode
+        if (!singlePage) {
+            outName = path.absolutePath
+        }
     }
 
+    fun save() {
+        val document = if (singlePage) {
+            // End the singlePage'd document
+            HtmlToLatexWriter(content.toString(), true).run {
+                commit()
+                endDocument()
+                toString()
+            }
+        } else {
+            content.toString()
+        }
+
+        // No ned to save an empty document
+        if (document.isEmpty()) {
+            return
+        }
+
+        val outDir = File("out", if (singlePage) "" else File(outName).parent)
+
+        if (!outDir.exists()) {
+            outDir.mkdirs()
+        }
+
+        // Save the LaTeX document
+        val file = File("out", outName.replace(".html", ".tex"))
+
+        val writer = FileWriter(file, false)
+
+        Log.i("${file.path} created")
+
+        writer.write(document)
+        writer.flush()
+        writer.close()
+    }
 }

bonus2/src/ch/usi/inf/atelier/group1/Main.kt

@@ -3,14 +3,25 @@
  */
 package ch.usi.inf.atelier.group1
 
+import ch.usi.inf.atelier.group1.util.Log
 import java.io.File
 
 object Main {
+    private lateinit var parser: HtmlParser
 
     @JvmStatic
     fun main(args: Array<String>) {
 
+        if (args.isEmpty()) {
+            Log.e("At least one argument is needed")
+            return
+        }
+
+        parser = HtmlParser(arrayOf("-s", "--singlepage").contains(args[0]))
+
         args.forEach(this::convert)
+
+        parser.save()
     }
 
     /**
@@ -19,10 +30,15 @@ object Main {
      * @param path of the file (or directory containing files) to be converted
      */
     private fun convert(path: String) {
+        // Ignore the flags
+        if (path.startsWith("-")) {
+            return
+        }
+
         val file = File(path)
 
         if (!file.isDirectory) {
-            HtmlParser(file).parse()
+            parser.parse(file)
             return
         }
 

bonus2/src/ch/usi/inf/atelier/group1/jekyll/HtmlToLatexWriter.kt

@@ -7,7 +7,7 @@ import org.jsoup.Jsoup
 import org.jsoup.parser.Parser
 import java.util.regex.Pattern
 
-class HtmlToLatexWriter(private var content: String) {
+class HtmlToLatexWriter(private var content: String, private val singlePage: Boolean) {
     private val document = StringBuilder()
 
     /**
@@ -30,11 +30,18 @@ class HtmlToLatexWriter(private var content: String) {
         insert("\\tableofcontents\\")
     }
 
+    /**
+     * Store the converted html text into
+     * the writer content so it can be later exported
+     */
+    fun commit() {
+        insert(content)
+    }
+
     /**
      * End the document
      */
     fun endDocument() {
-        insert(content)
         insert("\\end{document}", afterLine = true)
     }
 
@@ -116,26 +123,36 @@ class HtmlToLatexWriter(private var content: String) {
                 .replaceTag("<h5>", "</h5>", "", "")
                 .replaceTag("<p>", "</p>", "\n", "")
                 .replaceTag("<a>", "</a>", "", "")
+
+        if (singlePage) {
+            content = content.replaceTag("<h3>", "</h3>", "", "")
+        }
     }
 
     /**
      * Replace <h1> with LaTeX \section
      */
     fun changeSection() {
-        content = content.replaceTag("<h1>", "</h1>", "\\section{", "}\n")
+        content = content.replaceTag("<h1>", "</h1>",
+                if (singlePage) "\\subsection{" else "\\section{", "}\n")
     }
 
     /**
      * Replace <h2> with LaTeX \subsection
      */
     fun changeSubSection() {
-        content = content.replaceTag("<h2>", "</h2>", "\\subsection{", "}\n")
+        content = content.replaceTag("<h2>", "</h2>",
+                if (singlePage) "\\subsubsection{" else "\\subsection{", "}\n")
     }
 
     /**
      * Replace <h3> with LaTeX \subsubsection
      */
     fun changeSubSubSection() {
+        if (singlePage) {
+            return
+        }
+
         content = content.replaceTag("<h3>", "</h3>", "\\subsubsection{", "}\n")
     }
 
@@ -197,7 +214,7 @@ class HtmlToLatexWriter(private var content: String) {
      * @param author the text being inserted
      */
     fun addAuthor(author: String) {
-        insert(AUTHOR.format(author), true, true)
+        insert(if (singlePage) "\\large $author \\normalsize\\\\" else AUTHOR.format(author), true, true)
     }
 
     /**
@@ -206,7 +223,19 @@ class HtmlToLatexWriter(private var content: String) {
      * @param title the text being inserted
      */
     fun addTitle(title: String) {
+        insert(if (singlePage) "\\section{$title}\n" else TITLE.format(title), true, true)
+    }
+
+    /**
+     * Add an header for the html file in singlePage mode
+     */
+    fun addSinglePageInfo(title: String, author: String) {
+        if (!singlePage) {
+            return
+        }
+
         insert(TITLE.format(title), true, true)
+        insert(AUTHOR.format(author), false, true)
     }
 
     /**

bonus2/src/ch/usi/inf/atelier/group1/util/Log.kt

@@ -8,14 +8,45 @@ import java.util.*
 
 object Log {
 
-    fun e(exception: Exception) {
-        print('E', exception.message ?: "Unknown error", true)
+    /**
+     * Log an exception as an error
+     *
+     * @param message The log message
+     * @param shouldThrow Whether the exception should be thrown
+     */
+    fun <T : Exception> e(exception: T, shouldThrow: Boolean) {
+        e(exception.message ?: "Unknown error")
+
+        if (shouldThrow) {
+            throw exception
+        }
     }
 
+    /**
+     * Log an error
+     *
+     * @param message The log message
+     */
+    fun e(message: String) {
+        print('E', message , true)
+    }
+
+    /**
+     * Log an object using its toString() method.
+     *
+     * @param obj The object which content will be printed.
+     */
     fun i(obj: Any) {
         print('I', obj.toString(), false)
     }
 
+    /**
+     * Print a log with a date and prefix
+     *
+     * @param prefix The prefix of the log. Helps differentiating the various types of logs
+     * @param message The message that will be displayed in the log
+     * @param isErr Whether the log should be printed as an error
+     */
     private fun print(prefix: Char, message: String, isErr: Boolean) {
         val time = SimpleDateFormat("yyyy-MM-dd hh:mm").format(Date())
 

bonus2/src/ch/usi/inf/atelier/group1/util/extensions/File.kt

@@ -26,7 +26,7 @@ fun File.getContent(): String {
             line = reader.readLine()
         }
     } catch (e: IOException) {
-        Log.e(exception = e)
+        Log.e(e, true)
     }
 
     return content.toString()

bonus2/src/ch/usi/inf/atelier/group1/util/extensions/TexDocument.kt

@@ -11,6 +11,8 @@ import java.io.FileWriter
 
 /**
  * Insert author and title from Jekyll file
+ *
+ * @param file File from which the author will be read
  */
 fun HtmlToLatexWriter.insertJekyllHeader(file: JekyllPage) {
     val title = file.header["title"] ?: "Unknown title"
@@ -18,22 +20,4 @@ fun HtmlToLatexWriter.insertJekyllHeader(file: JekyllPage) {
 
     addTitle(title)
     addAuthor(author)
-}
-
-fun HtmlToLatexWriter.writeTo(path: String) {
-    val outDir = File("out", File(path).parent)
-
-    if (!outDir.exists()) {
-        outDir.mkdirs()
-    }
-
-    val file = File("out", path)
-
-    val writer = FileWriter(file, false)
-
-    Log.i("${file.path} created")
-
-    writer.write(toString())
-    writer.flush()
-    writer.close()
 }