Remove dependency on c++ regex library

gangeli · Dec 1, 2015 · 1d3d10a · 1d3d10a
1 parent 2ab425b
commit 1d3d10a
Show file tree

Hide file tree

Showing 9 changed files with 143 additions and 157 deletions.
diff --git a/etc/mkGraph.sh b/etc/mkGraph.sh
@@ -128,7 +128,7 @@ PRIVATIVE_WORDS=`echo "^(" \
      "proposed" \
      ")	[0-9]+" | sed -e 's/| /|/g'`
 
-cat "$DIR/$GRAPH_DATA/edge_*.txt" |\
+cat "$DIR/$GRAPH_DATA/"edge_*.txt |\
   sed -e 's/_/ /g' |\
   awk -F'	' '{ print $1 "\t" $2 }' |\
   egrep "$PRIVATIVE_WORDS" |\

diff --git a/pub/acl2015/talk/clause.tex b/pub/acl2015/talk/clause.tex
@@ -253,7 +253,3 @@
   \item POS tag of governor; dependent.
 \end{itemize}
 \end{frame}
-
-
-
-
diff --git a/pub/ai2/motivation.tex b/pub/ai2/motivation.tex
@@ -48,12 +48,8 @@
   \item Grounding 3,800 rules takes 7 hours (Alchemy).
   \pause
   \item Add Chris R\'{e} $+$ decades of DB research: 106 seconds.
-  \item $\dots$ but relation extraction over 1M sentences needs 500GB memory.
+  \item $\dots$ but still needs lots of memory.
 \end{itemize}
-\vspace{1ex}
-\pause
-
-\hh{We want to run inference over 300M premises}
 
 \end{frame}
 
@@ -102,7 +98,6 @@
 \pause
 \item But, can still infer: \w{Most students learned it at a school.}
 \end{itemize}
-
 \end{frame}
 
 
diff --git a/pub/figures.tex b/pub/figures.tex
@@ -1,4 +1,5 @@
 
+
 %
 % -----
 % The full Finite State Automata over inference states
@@ -710,28 +711,28 @@
 
 \def\exampleStepOne{\exampleTree{ROOT \\ $~$}
                                 {No carnivores eat animals}
-                                {\textbf{The} carnivores eat animals}
-                                {No \textbf{cats} eat animals}}
+                                {\darkblue{The} carnivores eat animals}
+                                {No \darkblue{animals} eat animals}}
 \def\exampleStepTwo{\exampleTree{No carnivores eat animals}
                                 {The carnivores eat animals}
-                                {The \textbf{feline} eats animals}
-                                {\textbf{All} carnivores eat animals}}
+                                {The \darkblue{feline} eats animals}
+                                {\darkblue{All} carnivores eat animals}}
 \def\exampleStepThree{\exampleTree{The carnivores eat animals}
                                   {The feline eats animals}
-                                  {The \textbf{cat} eats animals}
-                                  {The cat eats \textbf{chordate}}}
+                                  {The \darkblue{cat} eats animals}
+                                  {The cat eats \darkblue{chordate}}}
 \def\exampleStepFour{\exampleTree{The feline eats animals}
                                  {The cat eats animals}
-                                 {The cat eats \textbf{chordates}}
-                                 {The \textbf{kitty} eats animals}}
+                                 {The cat eats \darkblue{chordates}}
+                                 {The \darkblue{kitty} eats animals}}
 \def\exampleStepFive{\exampleTree{The cat eats animals}
                                  {The cat eats chordates}
-                                 {The cat eats \textbf{mice}}
-                                 {The cat eats \textbf{dogs}}}
+                                 {The cat eats \darkblue{mice}}
+                                 {The cat eats \darkblue{dogs}}}
 \def\exampleStepSix{\exampleTreeEnd{The cat eats chordates}
                                    {The cat eats mice}
-                                   {The cat ate \textbf{a} mouse}
-                                   {The \textbf{kitty} eats mice}}
+                                   {The cat ate \darkblue{a} mouse}
+                                   {The \darkblue{kitty} eats mice}}
 \def\exampleStepEnd{
   \setstyles
   \begin{tikzpicture}[grow=down, sloped]
@@ -859,90 +860,6 @@
 
 
 
-%
-% -----
-% MIML-RE Figures
-% -----
-%
-
-\def\mimlStyles{
-  \tikzstyle{var}=[circle,thick,draw=black!75,fill=black!0,minimum size=10mm]
-  \tikzstyle{ovar}=[circle,thick,draw=black!75,fill=gray!20,minimum size=10mm]
-}
-
-\newcommand\dsPlate{
-\begin{tikzpicture}[->,>=stealth,shorten >=1pt,auto,node distance=1.5cm and 0.75cm,
-                    semithick]
-  \mimlStyles
-  \node[ovar]  (y)      []                     {$y$};
-  \node[ovar] (x)      [below =of y]          {$x$};
-
-  \path (x) edge (y);
-\end{tikzpicture}
-}
-
-\newcommand\miPlate{
-\begin{tikzpicture}[->,>=stealth,shorten >=1pt,auto,node distance=1.5cm and 0.75cm,
-                    semithick]
-  \mimlStyles
-  \node[ovar] (y)      []                        {$y$};
-
-  \node[var]  (zTwo)      [below =of y]          {$z_2$};
-  \node[ovar] (xTwo)      [below =of zTwo]       {$x_2$};
-
-  \node[var]  (zOne)      [left =of zTwo]        {$z_1$};
-  \node[ovar] (xOne)      [left =of xTwo]        {$x_1$};
-
-  \node[var]  (zThree)    [right =of zTwo]       {$z_3$};
-  \node[ovar] (xThree)    [right =of xTwo]       {$x_3$};
-
-  \path (xOne) edge (zOne)
-        (xTwo) edge (zTwo)
-        (xThree) edge (zThree)
-        (zOne) edge (y)
-        (zTwo) edge (y)
-        (zThree) edge (y);
-\end{tikzpicture}
-}
-
-\newcommand\mimlPlate{
-\begin{tikzpicture}[->,>=stealth,shorten >=1pt,auto,node distance=1.5cm and 0.75cm,
-                    semithick]
-  \mimlStyles
-  \node[ovar] (yOne)                          {$y_1$};
-  \node[ovar] (yTwo)   [right =of yOne]       {$y_2$};
-  \node[]     (yDots)  [right =of yTwo]       {$\dots$};
-  \node[ovar] (yNMin)  [right =of yDots]      {$y_{n-1}$};
-  \node[ovar] (yN)     [right =of yNMin]      {$y_n$};
-
-  \node[var]  (zTwo)      [below =of yDots]   {$z_2$};
-  \node[ovar] (xTwo)      [below =of zTwo]    {$x_2$};
-
-  \node[var]  (zOne)      [left =of zTwo]     {$z_1$};
-  \node[ovar] (xOne)      [left =of xTwo]     {$x_1$};
-
-  \node[var]  (zThree)    [right =of zTwo]    {$z_3$};
-  \node[ovar] (xThree)    [right =of xTwo]    {$x_3$};
-
-  \path (xOne) edge (zOne)
-        (xTwo) edge (zTwo)
-        (xThree) edge (zThree)
-        (zOne) edge (yOne)
-               edge (yTwo)
-               edge (yNMin)
-               edge (yN)
-        (zTwo) edge (yOne)
-               edge (yTwo)
-               edge (yNMin)
-               edge (yN)
-        (zThree) edge (yOne)
-                 edge (yTwo)
-                 edge (yNMin)
-                 edge (yN);
-\end{tikzpicture}
-}
-
-
 %
 % Standalone figures
 %

diff --git a/pub/macros.tex b/pub/macros.tex
@@ -17,9 +17,6 @@
 \newcommand\sys[1]{\textsc{#1}}
 \newcommand\w[1]{\textit{\darkgreen{#1}}}
 \newcommand\ww[1]{\textit{#1}}
-\newcommand\true[1]{\darkgreen{\textit{#1}}}
-\newcommand\false[1]{\darkred{\textit{#1}}}
-\newcommand\unknown[1]{\orange{\textit{#1}}}
 \newcommand{\indentitem}{\setlength\itemindent{25pt}}
 
 \newcommand\denote[1]{\ensuremath{\llbracket#1\rrbracket}}
@@ -54,6 +51,10 @@
 \newcommand{\noop}{\textcolor{white}{NOOP}}
 \newcommand{\noopTab}{\begin{tabular}{c} \textcolor{white}{NOOP} \\ \textcolor{white}{NOOP} \end{tabular}}
 
+\newcommand\true[1]{\darkgreen{\checkmark\textit{#1}}}
+\newcommand\false[1]{\darkred{\xmark$~\,$\textit{#1}}}
+\newcommand\unknown[1]{?\orange{\textit{#1}}}
+
 \newcommand{\verticalcenter}[1]{\begingroup
 \setbox0=\hbox{#1}%
 \parbox{\wd0}{\box0}\endgroup}
@@ -78,11 +79,10 @@
 % KBP Specific
 % An entity
 \def\ent#1{\text{\small{\textsc{#1}}}}
-% A phrase
-\def\w#1{\textit{#1}}
 
 % An extraction, e.g., "Obama born_in Hawaii"
 \newcommand\extr[3]{\mbox{\ent{#1}\ $~$\rel{#2}\ $~$\ent{#3}}}
+\newcommand\triple[3]{(\mbox{\ent{#1}; $~$\rel{#2}; $~$\ent{#3}})}
 % A clause in a logical form, e.g., "born_in(Obama, Hawaii)"
 \newcommand\clause[3]{\mbox{\rel{#2}\ensuremath{(#1, #3)}}}
 
@@ -109,3 +109,11 @@
 \end{tabular}
 \vspace{2mm}
 }
+
+\tikzset{
+    invisible/.style={opacity=0},
+    visible on/.style={alt=#1{}{invisible}},
+    alt/.code args={<#1>#2#3}{%
+      \alt<#1>{\pgfkeysalso{#2}}{\pgfkeysalso{#3}} % \pgfkeysalso doesn't change the path
+    },
+  }
diff --git a/src/JavaBridge.cc b/src/JavaBridge.cc
@@ -51,6 +51,7 @@ JavaBridge::JavaBridge() {
     snprintf(classpath, 1024, "%s/naturalli_preprocess.jar:%s/jaws.jar:%s/../lib/jaws.jar:%s:%s", 
         thisDir.c_str(), thisDir.c_str(), thisDir.c_str(),
         CORENLP, CORENLP_MODELS);
+    printf("%s\n", classpath);
 
     // Start program
     dup2(pipeIn[0], STDIN_FILENO);

diff --git a/src/NaturalLIIO.cc b/src/NaturalLIIO.cc
@@ -16,7 +16,7 @@
 #include <sys/types.h>
 #include <sys/resource.h>
 #include <signal.h>
-#include <regex>
+#include <regex.h>
 #include <thread>
 #include <unistd.h>
 
@@ -208,31 +208,51 @@ bool to_bool(std::string str) {
   return b;
 }
 
-/** The regex for a key@specifier=value triple */
-regex regexSetValue("([^ ]+) *@ *([^ ]+) *= *([^ ]+)",
-                    std::regex_constants::egrep);
-/** The regex for a key=value pair */
-regex regexSetFlag("([^ ]+) *= *([^ ]+) *", std::regex_constants::egrep);
 
+//
+// Get the gloss of a regex group matched against a string.
+//
+string regexGroup(const regmatch_t& pmatch, const string& source) {
+  uint32_t start = pmatch.rm_so;
+  uint32_t end   = pmatch.rm_eo;
+  char group[end - start + 1];
+  memcpy(group, source.c_str() + start, end - start);
+  group[end - start] = 0;
+  return string(group);
+}
 
 //
 // parseMetadata()
 //
 bool parseMetadata(const char *rawLine, SynSearchCosts **costs,
                    vector<AlignmentSimilarity>* alignments,
                    syn_search_options *opts) {
+  // Compile the regexes
+  regex_t regexSetValue;
+  if (regcomp(&regexSetValue, "([^ ]+) *@ *([^ ]+) *= *([^ ]+)", REG_EXTENDED)) {
+    fprintf(stderr, "Could not compile regex set_value");
+    exit(1);
+  }
+  regex_t regexSetFlag;
+  if (regcomp(&regexSetFlag, "([^ ]+) *= *([^ ]+) *", REG_EXTENDED)) {
+    fprintf(stderr, "Could not compile regex set_flag");
+    exit(1);
+  }
+  regmatch_t pmatch[4];
+
+  // Error checks
   assert(rawLine[0] != '\0');
   assert(rawLine[0] == '%');
+
+  // Run regex
   string line(&(rawLine[1]));
-  smatch result;
-  if (regex_search(line, result, regexSetValue)) {
-    if (result.size() != 4) {
-      return false;
-    }
-    string toSet = result[1].str();
-    string key = result[2].str();
-    string value = result[3].str();
-    uint32_t ivalue = atoi(result[3].str().c_str());
+  if (!regexec(&regexSetValue, line.c_str(), 4, pmatch, 0)) {
+    string toSet = regexGroup(pmatch[1], line);
+    string key = regexGroup(pmatch[2], line);
+    string value = regexGroup(pmatch[3], line);
+
+
+    uint32_t ivalue = atoi(value.c_str());
     if (toSet == "mutationLexicalCost") {
       (*costs)->mutationLexicalCost[indexEdgeType(key)] = atof(value.c_str());
     } else if (toSet == "insertionLexicalCost") {
@@ -243,12 +263,16 @@ bool parseMetadata(const char *rawLine, SynSearchCosts **costs,
       (*costs)->transitionCostFromFalse[indexNatlogRelation(key)] = atof(value.c_str());
     } else {
       fprintf(stderr, "Unknown key: '%s'\n", toSet.c_str());
+      regfree(&regexSetValue);
+      regfree(&regexSetFlag);
       return false;
     }
+    regfree(&regexSetValue);
+    regfree(&regexSetFlag);
     return true;
-  } else if (regex_search(line, result, regexSetFlag)) {
-    string toSet = result[1].str();
-    string value = result[2].str();
+  } else if (!regexec(&regexSetFlag, line.c_str(), 3, pmatch, 0)) {
+    string toSet = regexGroup(pmatch[1], line);
+    string value = regexGroup(pmatch[2], line);
     if (toSet == "maxTicks") {
       opts->maxTicks = atof(value.c_str());
       fprintf(stderr, "set maxTicks to %u\n", opts->maxTicks);
@@ -284,12 +308,18 @@ bool parseMetadata(const char *rawLine, SynSearchCosts **costs,
       }
     } else {
       fprintf(stderr, "Unknown flag: '%s'\n", toSet.c_str());
+      regfree(&regexSetValue);
+      regfree(&regexSetFlag);
       return false;
     }
+    regfree(&regexSetValue);
+    regfree(&regexSetFlag);
     return true;
   }
   fprintf(stderr, "WARNING line NOT parsed as a directive: '%s'\n",
           line.c_str());
+  regfree(&regexSetValue);
+  regfree(&regexSetFlag);
   return false; // By default, no metadata
 }
 

diff --git a/src/edu/stanford/nlp/naturalli/entail/NaturalLIClassifier.java b/src/edu/stanford/nlp/naturalli/entail/NaturalLIClassifier.java
@@ -37,8 +37,8 @@ public class NaturalLIClassifier implements EntailmentClassifier {
   public static boolean USE_LUCENE = true;
 
   @Execution.Option(name="naturalli.weight", gloss="The weight to incorporate NaturalLI with")
-  public static double ALIGNMENT_WEIGHT = 6.0;
-  public static double CLASSIFIER_WEIGHT = 1.0 / 4.0;
+  public static double ALIGNMENT_WEIGHT = 0.5;
+  public static double CLASSIFIER_WEIGHT = 0.5;
 
   @Execution.Option(name="naturalli.incache", gloss="The cache to read from")
   private static String NATURALLI_INCACHE = "logs/train_all.cache";