phd_thesis_template/generator.php

<?php
  global $options;
  global $preamble;
  $preamble = "";
  global $existing_commands;
  global $included_usepackages;
  $included_usepackages = [];
  global $additional_usepackages;
  $additional_usepackages = [];
  $steps = 10;
  function user_consent($msg,$defaultyes = null,$defaultno = null)
  {
    global $options;
    if (isset($options["-n"]))
      return false;
    if ($defaultyes !== null && isset($options[$defaultyes]))
      return true;
    if ($defaultno !== null && isset($options[$defaultno]))
      return false;
    $response = false;
    do
    {
      echo "$msg (y/n)\n";
      $response = rtrim(fgets(STDIN));
      if ($response === "n" || $response == "N")
        return false;
    } while ($response !== "y" && $response !== "Y" && $response !== "");
    return true;
  }
  function user_prompt($msg,$default)
  {
    global $options;
    echo "$msg [Default: $default]\n";
    $response = "";
    if (!isset($options["-n"]))
      $response = rtrim(fgets(STDIN));
    if ($response == "")
      $response = $default;
    echo "You entered: '$response'\n";
    return $response;
  }
  function print_help()
  {
    $name = basename(__FILE__);
    echo<<<END
Usage: php $name /path/folder1/main.tex /path/folder2/article.tex ... /path/folderN/paper.tex\n
Generate a thesis template from the provided paper tex files and folders.

The absolute minimum for a cumulative thesis is currently 3 papers first-authored and 6 in total.

--force-overwrite        overwrite files and folders
--no-overwrite           do not overwrite files and folders
--no-compile-check       do not perform a compile check
--use-backups            restore backups before making changes
--not-sloppy             disable use of sloppypar around places that struggle without
                         (mainly the bibliographies)
--cv                     add a CV
                         (unusual for PhD theses, required for habilitation theses)
--no-stat-declaration    do not add the statutory declaration
                         (required for PhD thesis, unusual for habilitation theses)
-n                       skip all prompts (implies --no-overwrite)
--help                   display this help and exit
END;
    exit(0);
  }
  function error_exit($msg,$usage = false)
  {
    echo "ERROR: $msg\n";
    if ($usage)
      echo "Usage: php ".basename(__FILE__)." /path/folder1/main.tex /path/folder2/article.tex ... /path/folderN/paper.tex\n";
    exit(-1);
  }
  function check_software()
  {
    if (!str_starts_with(shell_exec("latexmk --version"),"Latexmk,"))
      error_exit("latexmk not installed");
    if (!str_starts_with(shell_exec("pdftotext --help 2>&1"),"pdftotext version"))
      error_exit("pdftotext not installed");
  }
  function rrmdir($dir) {
    if (is_dir($dir)) {
      $files = scandir($dir);
      foreach ($files as $file)
      if ($file != "." && $file != "..") rrmdir("$dir/$file");
      rmdir($dir);
    }
    else if (file_exists($dir)) unlink($dir);
  }
  function rcopy($src, $dst) {
    if (file_exists($dst)) rrmdir($dst);
    if (is_dir($src)) {
      mkdir($dst);
      $files = scandir($src);
      foreach ($files as $file)
      if ($file != "." && $file != "..") rcopy("$src/$file", "$dst/$file"); 
    }
    else if (file_exists($src)) copy($src, $dst);
  }
  function rscandir($dir,$extensions = []) {
    $result = [];
    if (!str_ends_with($dir,"/"))
      $dir .= "/";
    if (str_ends_with($dir,"latex.out/"))
      return $result;
    $array = scandir($dir);
    foreach ($array as $item) {
      if (!str_starts_with($item,"."))
      {
        if (!is_dir($dir.$item))
        {
          $extension_match = true;
          if ($extensions != [])
          {
            $extension_match = false;
            foreach ($extensions as $e)
            {
              if (str_ends_with($item,"$e"))
              {
                $extension_match = true;
              }
            }
          }
          if ($extension_match)
            $result[] = $dir.$item;
        }
        else
          $result = array_merge($result, rscandir($dir.$item,$extensions));
      }
    }
    return $result;
  }
  function unique_targets($targets)
  {
    $parts_required = 1;
    $unique_targets = [];
    while (count($unique_targets) != count($targets))
    {
      foreach ($targets as $t)
      {
        $tparts = explode("/",$t);
        $tpcount = count($tparts);
        $tparts = array_splice($tparts,$tpcount-$parts_required,$parts_required);
        $t = implode("_",$tparts);
        if (isset($unique_targets[$t]))
        {
          $parts_required++;
          $unique_targets = [];
          break;
        }
        $unique_targets[$t] = 1;
      }
    }
    return array_keys($unique_targets);
  }
  function check_folders($argv)
  {
    $args = array_slice($argv,1);
    $sourcedirs = [];
    foreach ($args as $a)
    {
      if (!str_starts_with($a,"-"))
        $sourcedirs[] = $a;
    }
    if (count($sourcedirs) < 1)
    {
      error_exit("Too few papers. Need at least one paper to run this.",true);
    }
    if (count($sourcedirs) < 3)
    {
      echo "Warning: Too few papers. The statutes require as an absolute minimum for a cumulative thesis at least 3 first-author papers and at least 6 in total.";
    }
    if (count($sourcedirs) < 5)
    {
      echo "Warning: Few papers. Check your institute guidelines to see how many papers are recommended. Also the statutes require as an absolute minimum for a cumulative thesis at least 6 papers in total.";
    }
    $texfiles = [];
    for ($i = 0; $i < count($sourcedirs); $i++)
    {
      $texfiles[$i] = basename($sourcedirs[$i]);
      $sourcedirs[$i] = dirname($sourcedirs[$i]);
    }
    $targets = unique_targets($sourcedirs);
    return [$targets,$texfiles,$sourcedirs];
  }
  function check_and_copy_folders($argv)
  {
    [$targets,$texfiles,$sourcedirs] = check_folders($argv);
    //echo "Source -> Target Directory Mapping:\n";
    $mapping = array_combine($sourcedirs,$targets);
    //print_r($mapping);
    foreach ($mapping as $s => $t)
    {
      if (!file_exists($s))
        error_exit("$s not found");
      if (file_exists($t) && !user_consent("$t already exists... overwrite?","--force-overwrite","--no-overwrite"))
      {
        echo "Skipping ".$s.", ".$t." already exists...\n";
      }
      else
      {
        echo "Copying ".$s." to ".$t."...\n";
        rcopy($s,$t);
      }
    }
    return array_combine($targets,$texfiles);
  }
  function file_get_and_backup($f,$bak = ".orig_bak")
  {
    global $options;
    if (isset($options["--use-backups"]))
    {
      if (file_exists("$f$bak"))
        copy("$f$bak", $f);
    }
    $content = file_get_contents("$f");
    if (!file_exists("$f$bak"))
      file_put_contents("$f$bak",$content);
    return $content;
  }
  function adjust_references($papers,$extensions = [])
  {
    $bibresources = [];
    foreach ($papers as $d => $f)
    {
      $files = rscandir($d,$extensions);
      foreach ($files as $file)
      {
        if (str_ends_with($file,".bib"))
        {
          $bib = file_get_and_backup($file);
          $bib = preg_replace("/(^\s*@\s*[a-z]+\s*\{\s*)((?!$d)\S)/i",'${1}'.$d.':${2}',$bib);
          $bib = preg_replace("/(\n\s*@\s*[a-z]+\s*\{\s*)((?!$d)\S)/i",'${1}'.$d.':${2}',$bib);
          file_put_contents($file,$bib);
          $bibresources[$d] = $file;
        }
        else
        {
          $tex = file_get_and_backup($file);
          // labels of lstlistings etc
          $tex = preg_replace("/([\[,]\s*label\s*=\s*)((?!$d)[^\[,]+[\[,])/i",'${1}'.$d.':${2}',$tex);
          // normal labels
          $tex = preg_replace("/(\\\label\s*\{\s*)((?!$d)[^}]+\})/i",'${1}'.$d.':${2}',$tex);
          // cref ref autoref
          $refs = [];
          preg_match_all("/\\\(cref|ref|autoref)\s*\{\s*[^}]+\}/i",$tex,$refs);
          foreach ($refs[0] as $ref)
          {
            $oldref = $ref;
            $ref = preg_replace("/([{,])\s*((?!$d)[^,}]+[},])/i",'${1}'.$d.':${2}',$ref);
            $ref = preg_replace("/([{,])\s*((?!$d)[^,}]+[},])/i",'${1}'.$d.':${2}',$ref);
            $tex = str_replace($oldref,$ref,$tex);
          }
          // cite citeA citeauthor etc
          $cites = [];
          preg_match_all("/\\\([a-z]*cite[a-z]*)\s*\{\s*[^}]+\}/i",$tex,$cites);
          foreach ($cites[0] as $cite)
          {
            $oldcite = $cite;
            $cite = preg_replace("/([{,])\s*((?!$d)[^,}]+[},])/i",'${1}'.$d.':${2}',$cite);
            $cite = preg_replace("/([{,])\s*((?!$d)[^,}]+[},])/i",'${1}'.$d.':${2}',$cite);
            $tex = str_replace($oldcite,$cite,$tex);
          }
          file_put_contents($file,$tex);
        }
      }
    }
    return $bibresources;
  }
  function stripformatting($tex)
  {
    return str_ireplace(['\Large ','\bf ','\rm ','\\\\'],[' ',' ',' ',' '],$tex);
  }
  function getusepackages($tex)
  {
    $packages = [];
    preg_match_all("/\n\s*(\\\\usepackage([^{}]*)(?<R>{((?:[^{}]+|(?&R))*)}))/i",$tex,$packages);
    if (isset($packages[1]))
    {
      return [$packages[1],$packages[4]];
    }
    return [[],[]];
  }
  function getmathops($tex)
  {
    $mathops = [];
    preg_match_all("/\n\s*(\\\\DeclareMathOperator([^{}]*)(?<R>{((?:[^{}]+|(?&R))*)})([^{}]*)(?<P>{((?:[^{}]+|(?&P))*)}))/i",$tex,$mathops);
    if (isset($mathops[4]))
      return [$mathops[1],$mathops[4]];
    return [];
  }
  function getnewcommands($tex)
  {
    $newcommands = [];
    preg_match_all("/\n\s*(\\\\newcommand([^{}]*)(?<R>{((?:[^{}]+|(?&R))*)})([^{}]*)(?<P>{((?:[^{}]+|(?&P))*)}))/i",$tex,$newcommands);
    if (isset($newcommands[4]))
      return $newcommands[4];
    return [];
  }
  function adjust_papers($papers,$extensions = [])
  {
    foreach ($papers as $d => $f)
    {
      $paper_fulltitle = "";
      $authors = "";
      $files = rscandir($d,$extensions);
      foreach ($files as $file)
      {
        if (!str_ends_with($file,"main.tex"))
          continue;
        $tex = file_get_and_backup($file,".interm_bak");
        if ($authors == "" && preg_match("/\n\s*\\\\author/i",$tex) == 1)
        {
          $authors = [];
          preg_match_all("/\n\s*\\\\author\s*(?<R>{((?:[^{}]+|(?&R))*)})/i",$tex,$authors);
          if (isset($authors[2]) && isset($authors[2][0]))
            $authors = trim(stripformatting($authors[2][0]));
        }
        if ($paper_fulltitle == "" && preg_match("/\n\s*\\\\title/i",$tex) == 1)
        {
          $title = [];
          preg_match_all("/\n\s*\\\\title\s*(?<R>{((?:[^{}]+|(?&R))*)})/i",$tex,$title);
          if (isset($title[2]) && isset($title[2][0]))
            $title = trim(stripformatting($title[2][0]));
          else
            $title = $d;
          
          $shorttitle = $title;
          if (strpos($title,":") !== false)
            $shorttitle = trim(explode(":",$title)[0]);
          $paper_fulltitle = trim(user_prompt("Please enter the full title for the paper in folder '$d'","$title"));
          $paper_shorttitle = trim(user_prompt("Please enter a short title for the paper in folder '$d'","$shorttitle"));
        }
        $header = "";
        if ($paper_fulltitle != "" && $authors != "")
        {
          $header=<<<END
\chapter[$paper_shorttitle]{$paper_fulltitle}\label{chapter:$d}
\section*{Publication Data}
\\fullcite{TODO}

\section*{Contributions}
TODO


\\newpage
\begin{center}
{\Large \\textbf{
$paper_fulltitle%
}}\\\\\
\\vspace{0.6cm}
{\large $authors%
} % TODO: check author list
\\end{center}
END;
        }
        $tex = preg_replace("/\\\documentclass.*\\n/i",'',$tex);
        $tex = preg_replace("/\\\begin{document}/i",$header,$tex);
        $tex = preg_replace("/\\\begin{abstract}/i",'\section*{Abstract}',$tex);
        $tex = preg_replace("/\\\end{abstract}/i",'',$tex);
        $tex = preg_replace("/\\\end{document}/i",'',$tex);
        $tex = preg_replace("/\n\s*\\\\author\s*(?<R>{((?:[^{}]+|(?&R))*)})/i","",$tex);
        $tex = preg_replace("/\n\s*\\\\title\s*(?<R>{((?:[^{}]+|(?&R))*)})/i","",$tex);
        $tex = preg_replace("/\n\s*\\\\date\s*(?<R>{((?:[^{}]+|(?&R))*)})/i","",$tex);
        $tex = preg_replace("/\n\s*\\\\maketitle/i","",$tex);
        [$ups,$ps] = getusepackages($tex);
        global $included_usepackages;
        global $additional_usepackages;
        foreach (array_diff($ps,$included_usepackages) as $k => $v) {
          if ($v != "usenix,epsfig,endnotes" && $v != "cite" && !($v == 'floatrow' && in_array('float',$included_usepackages)))
          {
            $additional_usepackages[] = $ups[$k];
            $included_usepackages[] = $v;
          }
        }
        $additional_usepackages = array_unique($additional_usepackages);
        sort($additional_usepackages);
        $tex = preg_replace("/\n\s*\\\\usepackage([^{}]*)(?<R>{((?:[^{}]+|(?&R))*)})/i","",$tex);
        
        $newcommands = getnewcommands($tex);
        global $existing_commands;
        $rmcommands = array_intersect($newcommands,$existing_commands);
        foreach ($rmcommands as $rmc)
        {
          $tex = preg_replace("/\n\s*\\\\newcommand([^{}]*)({\s*\\".$rmc."\s*}.*\n)/i","\n",$tex);
        }

        [$mathopsfull,$mathops] = getmathops($tex);
        global $existing_mathops;
        global $preamble;
        $addmathops = array_diff($mathops,$existing_mathops);
        foreach ($addmathops as $k => $v)
        {
          $preamble .= $mathopsfull[$k] . "\n";
          $existing_mathops[] = $v;
        }
        $rmmathops = array_intersect($mathops,$existing_mathops);
        foreach ($rmmathops as $rmo)
        {
          $tex = preg_replace("/\n\s*\\\\DeclareMathOperator([^{}]*)({\s*\\".$rmo."\s*}.*\n)/i","\n",$tex);
        }

        $matches = [];
        $tex = preg_replace("/\n\s*\\\\pgfplotsset\s*(?<R>{((?:[^{}]+|(?&R))*)})/i","",$tex);
        $tex = preg_replace("/\n\s*\\\\setlength{\s*\\\marginparwidth.*\n/i","\n",$tex);

        file_put_contents($file,$tex);
      }
    }
    return;
  }
  function check_options($argv)
  {
    $options = [];
    foreach ($argv as $a)
    {
      if (str_starts_with($a,"-"))
        $options[$a] = 1;
      if ($a == "-n")
        $options["--no-overwrite"] = 1;
    }
    return $options;
  }
  function compile_check($papers,$precopy = false)
  {
    if ($precopy)
      $papers = array_combine($papers,$papers);
    foreach ($papers as $p => $f)
    {
      echo "======= Paper $p -> $f START =======\n";
      if ($precopy)
      {
        echo "$p\n";
        if (strpos($p,"/") === false || str_starts_with($p,".") || !is_dir(dirname($p)))
          continue;
        $f = basename($p);
        $p = dirname($p);
      }
      else
      {
        shell_exec("cd $p; latexmk -c 2>/dev/null; latexmk -C 2>/dev/null; rm -f *.bbl");
      }
      $str = shell_exec("cd $p; latexmk -latexoption=\"-shell-escape\" -g -pdf $f 2>&1");
      $lines = explode("\n",$str);
      $skip = true;
      $failed = false;
      foreach($lines as $l)
      {
        if (str_starts_with($l,"Latexmk: ====List of undefined refs and citations:"))
          $skip = false;
        if ($skip)
          continue;
        if (stripos($l,"fail") !== false)
          $failed = true;
        echo "$l\n";
      }
      if ($failed || user_consent("Please check the PDF file. Are there any problems with invalid references to figures, tables or with the bibliography?","--","--no-overwrite"))
        error_exit("The paper did not compile properly, please fix the errors next before continuing.");
      echo "======= Paper $p -> $f  END  =======\n";
    }
  }
  check_software();
  $options = check_options($argv);
  if (isset($options["--help"]))
    print_help();
  echo "\n=== Step 1/$steps: Check Folders ===\n\n";
  [$targets,$texfiles,$sourcedirs] = check_folders($argv);
  echo "\n=== Step 2/$steps: Pre-Copy Compile Check ===\n\n";
  $targets_exist = true;
  foreach ($targets as $t)
    if (!is_dir($t))
      $targets_exist = false;
  if (isset($options["--no-compile-check"]) || $targets_exist)
    echo "Skipping...\n";
  else
    compile_check($argv,true);
  echo "\n=== Step 3/$steps: Copying Files ===\n\n";
  $papers = check_and_copy_folders($argv);
  echo "\n=== Step 4/$steps: Adjusting References (in *.bib *.tex *.tikz) ===\n\n";
  $bibresources = adjust_references($papers,[".bib",".tex",".tikz"]);
  echo "\n=== Step 5/$steps: Compile Check ===\n\n";
  if (isset($options["--no-compile-check"]))
    echo "Skipping...\n";
  else
    compile_check($papers);
  [$_t,$usepackages1] = getusepackages(file_get_contents("main.tex.php"));
  [$_t,$usepackages2] = getusepackages(file_get_contents("tugraz_defaults.sty"));
  global $included_usepackages;
  $included_usepackages = array_unique(array_merge($usepackages1, $usepackages2));
  sort($included_usepackages);
  $cmds1 = getnewcommands(file_get_contents("main.tex.php"));
  $cmds2 = getnewcommands(file_get_contents("tugraz_defaults.sty"));
  global $existing_commands;
  $existing_commands = array_unique(array_merge($cmds1, $cmds2));
  [$del,$mathops1] = getmathops(file_get_contents("main.tex.php"));
  [$del,$mathops2] = getmathops(file_get_contents("tugraz_defaults.sty"));
  global $existing_mathops;
  $existing_mathops = array_unique(array_merge($mathops1, $mathops2));
  echo "\n=== Step 5/$steps: Adjusting Papers (checking only *.tex files) ===\n\n";
  adjust_papers($papers,[".tex"]);
  echo "\n=== Step 6/$steps: Generate main.tex ===\n\n";
  $thesis_type = user_prompt("Enter Thesis Type ","PhD Thesis");
  $thesis_title = user_prompt("Enter Thesis Title ","Security of TODO");
  $thesis_part1_title = user_prompt("Enter Introductory Part Title ","Introduction to the Security of TODO");
  $thesis_author = user_prompt("Enter Your Name ","Harry Potter");
  $thesis_date = user_prompt("Enter Prospective Defense Month", "July 1998");
  $thesis_institute = user_prompt("Enter Your Institute","Institute for Applied Information Processing and Communications");
  $thesis_assessors = user_prompt("Enter Names of your Assessors (comma-separated)","Severus Snape, Minerva McGonagall");
  $num_publications_in_thesis = count($papers);
  $num_publications = user_prompt("How many publications did you co-author during your PhD? (6 is the absolute minimum for a cumulative thesis)","6");

  $sloppy_begin = "";
  $sloppy_end = "";
  if (!isset($options["--no-sloppy"]))
  {
    $sloppy_begin = '\begin{sloppypar}';
    $sloppy_end = '\end{sloppypar}';
  }

  ob_start();
  require "main.tex.php";
  $maintex_content = ob_get_contents();
  ob_end_clean();
  //if (!file_exists("main.tex") || user_consent("main.tex already exists. overwrite?")) // TODO: enable this check
  file_put_contents("main.tex",$maintex_content);

  if (!file_exists("mypreamble.sty"))
    file_put_contents("mypreamble.sty","\ProvidesPackage{mypreamble}\n\n".implode("\n",$additional_usepackages)."\n\n".$preamble);

  @mkdir("tikz");

  if (!file_exists("abstract.tex"))
    file_put_contents("abstract.tex","");
  if (!file_exists("intro.tex"))
    file_put_contents("intro.tex","");

  if (!file_exists("main.bib"))
    file_put_contents("main.bib","");
  if (isset($options["--cv"]))
  {
    if (!file_exists("cv.bib"))
      file_put_contents("cv.bib","");
    if (!file_exists("cv.tex"))
      file_put_contents("cv.tex","This is my CV! Thanks for checking it out!");
  }
?>