Make startup a lot faster

Summary: This greatly decreases startup time for the codebase by making the non-XHP fastpath a lot more sophisticated. One of the biggest fastpath false positive was markup in strings. The new fastpath correctly scans past strings which means we don't have to double parse nearly as many files. I implemented the fastpath in re2c which is the lightest weight scanner I could find. Basically the only thing it does is convert lists of regular expressions to huge state machines. Everything else you might expect from a scanner has to be implemented from scratch, which is just what I want. I implemented just enough of PHP's grammar to know when we're looking at actual code. Then I use the same XHP clues as I had in the old scanner. Reviewed By: dcorson Test Plan: Browsed around the lite site, etc. Revert: OK DiffCamp Revision: 69583
JosePeter · Oct 18, 2009 · dbaca37 · dbaca37
1 parent 2100aa3
commit dbaca37
Show file tree

Hide file tree

Showing 6 changed files with 109 additions and 44 deletions.
diff --git a/ext.hpp b/ext.hpp
@@ -4,7 +4,7 @@
 #endif
 #include "php.h"
 
-#define PHP_XHP_VERSION "1.3.2"
+#define PHP_XHP_VERSION "1.3.3"
 #define PHP_XHP_EXTNAME "xhp"
 
 extern zend_module_entry xhp_module_entry;

diff --git a/xhp/Makefile b/xhp/Makefile
@@ -11,7 +11,7 @@ endif
 all: libxhp.a libxhp.so
 
 clean:
-	-rm libxhp.a xhpize parser.yacc.cpp scanner.lex.cpp scanner.lex.hpp parser.yacc.output parser.yacc.hpp *.o 2>/dev/null
+	-rm libxhp.a xhpize parser.yacc.cpp scanner.lex.cpp scanner.lex.hpp parser.yacc.output parser.yacc.hpp fastpath.cpp *.o 2>/dev/null
 
 parser.yacc.cpp: parser.y
 	bison --debug --verbose -d -o $@ $<
@@ -24,17 +24,20 @@ scanner.lex.cpp: scanner.l
 
 scanner.lex.hpp: scanner.lex.cpp
 
+fastpath.cpp: fastpath.re
+	re2c -c -o $@ $<
+
 %.o: %.cpp
 	$(CXX) -c $(CPPFLAGS) -o $@ $<
 
 xhp_preprocess.o: xhp_preprocess.cpp scanner.lex.hpp parser.yacc.hpp
 parser.yacc.o: scanner.lex.hpp
 scanner.lex.o: parser.yacc.hpp
 
-libxhp.a: code_rope.o scanner.lex.o parser.yacc.o xhp_preprocess.o
+libxhp.a: code_rope.o scanner.lex.o parser.yacc.o fastpath.o xhp_preprocess.o
 	$(AR) -crs $@ $^
 
-libxhp.so: code_rope.o scanner.lex.o parser.yacc.o xhp_preprocess.o
+libxhp.so: code_rope.o scanner.lex.o parser.yacc.o fastpath.o xhp_preprocess.o
 	g++ -shared -Wl,-soname,libxhp.so -o libxhp.so $^
 
 xhpize: xhpize.cpp libxhp.a

diff --git a/xhp/fastpath.hpp b/xhp/fastpath.hpp
@@ -0,0 +1,4 @@
+#pragma once
+#include "xhp_preprocess.hpp"
+
+bool xhp_fastpath(const char* yy, const size_t len, const xhp_flags_t &flags);
diff --git a/xhp/fastpath.re b/xhp/fastpath.re
@@ -0,0 +1,93 @@
+#include "fastpath.hpp"
+#include <stdio.h>
+bool xhp_fastpath(const char* yy, const size_t len, const xhp_flags_t &flags) {
+  const char* eob = yy + len + 1;
+  const char* YYMARKER = NULL;
+  enum {
+    HTML,
+    PHP,
+    COMMENT_EOL,
+    COMMENT_BLOCK
+  } state = flags.eval ? PHP : HTML;
+
+  #define YYCURSOR yy
+  #define YYCTYPE char
+  #define YYGETCONDITION() state
+  #define YYFILL(ii) if (ii + YYCURSOR >= eob) return false
+
+  for (;;) {
+/*!re2c
+    re2c:condenumprefix = "";
+    re2c:yyfill:check = 0;
+
+    NEWLINE = ('\r'|'\n'|'\r\n');
+    WHITESPACE = [ \n\r\t]+;
+
+    <HTML> '<?php'([ \t]|NEWLINE) {
+      state = PHP;
+      continue;
+    }
+    <HTML> '<?='|'<?' {
+      if (flags.short_tags) {
+        state = PHP;
+      }
+      continue;
+    }
+    <HTML> '<%='|'<%' {
+      if (flags.asp_tags) {
+        state = PHP;
+      }
+      continue;
+    }
+    <HTML> [^] { continue; }
+
+    <PHP> '?>'|'</script'WHITESPACE*'>' {
+      state = HTML;
+      continue;
+    }
+    <PHP> '%>' {
+      if (flags.asp_tags) {
+        state = PHP;
+      }
+      continue;
+    }
+    <PHP> 'b'?'\''('\\'.|'\\\n'|[^\\']+)*'\''|
+          'b'?'\"'('\\'.|'\\\n'|[^\\"]+)*'\"' { continue; }
+    <PHP> '#'|'//' {
+      state = COMMENT_EOL;
+      continue;
+    }
+    <PHP> '/*' {
+      state = COMMENT_BLOCK;
+      continue;
+    }
+    <PHP> '::' { continue; }
+    <PHP> '</'|
+          '/>'|
+          ':'[a-zA-Z0-9]|
+          'element'|
+          ')'WHITESPACE*'[' {
+      return true;
+    }
+    <PHP> [^] { continue; }
+
+    <COMMENT_EOL> NEWLINE {
+      state = PHP;
+      continue;
+    }
+    <COMMENT_EOL> '?>' {
+      state = HTML;
+      continue;
+    }
+    <COMMENT_EOL> [^] { continue; }
+
+    <COMMENT_BLOCK> [^*] { continue; }
+    <COMMENT_BLOCK> '*/' {
+      state = PHP;
+      continue;
+    }
+    <COMMENT_BLOCK> [^] { continue; }
+*/
+  }
+  return false;
+}
diff --git a/xhp/xhp_preprocess.cpp b/xhp/xhp_preprocess.cpp
@@ -1,5 +1,6 @@
 #include "xhp.hpp"
 #include "xhp_preprocess.hpp"
+#include "fastpath.hpp"
 #include <sstream>
 using namespace std;
 extern int xhpdebug;
@@ -22,47 +23,11 @@ XHPResult xhp_preprocess(string &in, string &out, bool isEval, string &errDescri
   return xhp_preprocess(in, out, errDescription, errLineno, flags);
 }
 
-XHPResult xhp_preprocess(std::string &in, std::string &out, std::string &errDescription, uint32_t &errLineno, xhp_flags_t &flags) {
+XHPResult xhp_preprocess(std::string &in, std::string &out, std::string &errDescription, uint32_t &errLineno, const xhp_flags_t &flags) {
 
-  // Does this maybe contain XHP?
+  // Early bail if the code doesn't contain anything that looks like XHP
   char* buffer = const_cast<char*>(in.c_str());
-  bool maybe_xhp = false;
-  for (const char* jj = buffer; *jj; ++jj) {
-    if (*jj == '<') { // </a>
-      if (jj[1] == '/') {
-        maybe_xhp = true;
-        break;
-      }
-    } else if (*jj == '/') { // <a />
-      if (jj[1] == '>') {
-        maybe_xhp = true;
-        break;
-      }
-    } else if (*jj == ':') { // :fb:thing
-      if ((jj[1] >= 'a' && jj[1] <= 'z') ||
-          (jj[1] >= 'A' && jj[1] <= 'Z') ||
-          (jj[1] >= '0' && jj[1] <= '9')) {
-        maybe_xhp = true;
-        break;
-      } else if (jj[1] == ':') {
-        ++jj;
-      }
-    } else if (!memcmp(jj, "element", 7)) {
-      maybe_xhp = true;
-      break;
-    } else if (*jj == ')') { // foo()['etc']
-      do {
-        ++jj;
-      } while (*jj == ' ' || *jj == '\r' || *jj == '\n' || *jj == '\t');
-      if (*jj == '[') {
-        maybe_xhp = true;
-        break;
-      }
-    }
-  }
-
-  // Early bail
-  if (!maybe_xhp) {
+  if (!xhp_fastpath(buffer, in.length(), flags)) {
     return XHPDidNothing;
   }
 

diff --git a/xhp/xhp_preprocess.hpp b/xhp/xhp_preprocess.hpp
@@ -24,4 +24,4 @@ XHPResult xhp_preprocess(std::string &in, std::string &out, bool isEval,
 
 XHPResult xhp_preprocess(std::string &in, std::string &out,
                          std::string &errDescription, uint32_t &errLineno,
-                         xhp_flags_t &flags);
+                         const xhp_flags_t &flags);