manual/tools/texquote2: reimplement the tool in OCaml

manual/manual/Makefile

@@ -12,7 +12,7 @@ OCAMLDOC = $(if $(wildcard $(SRC)/ocamldoc/ocamldoc.opt),\
 # Copy and unprefix the standard library when needed
 include $(SRC)/ocamldoc/Makefile.unprefix
 
-TEXQUOTE = ../tools/texquote2
+TEXQUOTE = $(SRC)/byterun/ocamlrun ../tools/texquote2
 
 
 FILES = allfiles.tex biblio.tex foreword.tex version.tex warnings-help.etex

manual/manual/cmds/Makefile

@@ -7,7 +7,7 @@ TOOLS = ../../tools
 CAMLLATEX = $(SET_LD_PATH) \
   $(OCAMLRUN) $(TOOLS)/caml-tex2 \
   -caml "TERM=norepeat $(OCAML)" -n 80 -v false
-TEXQUOTE = $(TOOLS)/texquote2
+TEXQUOTE = $(OCAMLRUN) $(TOOLS)/texquote2
 TRANSF = $(SET_LD_PATH) $(OCAMLRUN) $(TOOLS)/transf
 
 

manual/manual/library/Makefile

@@ -15,7 +15,7 @@ OCAMLDOC = $(if $(wildcard $(CSLDIR)/ocamldoc/ocamldoc.opt),\
 include $(SRC)/ocamldoc/Makefile.unprefix
 
 
-TEXQUOTE = ../../tools/texquote2
+TEXQUOTE = $(SRC)/byterun/ocamlrun ../../tools/texquote2
 
 
 CORE_INTF = Pervasives.tex

manual/manual/refman/Makefile

@@ -7,7 +7,7 @@ TOOLS = ../../tools
 CAMLLATEX = $(SET_LD_PATH) \
   $(OCAMLRUN) $(TOOLS)/caml-tex2 \
   -caml "TERM=norepeat $(OCAML)" -n 80 -v false
-TEXQUOTE = $(TOOLS)/texquote2
+TEXQUOTE = $(OCAMLRUN) $(TOOLS)/texquote2
 TRANSF = $(SET_LD_PATH) $(OCAMLRUN) $(TOOLS)/transf
 
 

manual/manual/tutorials/Makefile

@@ -7,7 +7,7 @@ TOOLS = ../../tools
 CAMLLATEX = $(SET_LD_PATH) \
   $(OCAMLRUN) $(TOOLS)/caml-tex2 \
   -caml "TERM=norepeat $(OCAML)" -n 80 -v false
-TEXQUOTE = $(TOOLS)/texquote2
+TEXQUOTE = $(OCAMLRUN) $(TOOLS)/texquote2
 TRANSF = $(SET_LD_PATH) $(OCAMLRUN) $(TOOLS)/transf
 
 

manual/tools/Makefile

@@ -2,8 +2,6 @@ TOPDIR=../..
 COMPFLAGS=-I $(OTOPDIR)/otherlibs/str -I $(OTOPDIR)/otherlibs/unix
 include $(TOPDIR)/Makefile.tools
 
-OC_CFLAGS=-g -O
-
 all: texquote2 transf caml-tex2
 
 
@@ -17,6 +15,9 @@ caml-tex2: caml_tex2.ml
 	$(OCAMLC) $(TOPDIR)/compilerlibs/ocamlcommon.cma -I $(TOPDIR)/parsing \
 	  -o $@ str.cma unix.cma $<
 
+texquote2: texquote2.ml
+	$(OCAMLC) -o $@ $<
+
 
 %.cmo: %.ml
 	$(OCAMLC) -c $<
@@ -27,9 +28,6 @@ caml-tex2: caml_tex2.ml
 %.ml: %.mll
 	$(OCAMLLEX) $<
 
-%: %.c
-	$(CC) $(OC_CFLAGS) -o $@ $<
-
 
 .PHONY: clean
 clean:

manual/tools/texquote2.c

@@ -1,167 +0,0 @@
-#include <stdio.h>
-#include <ctype.h>
-
-char * transl[256];
-
-#define LINE_LENGTH 1024
-
-char line[LINE_LENGTH];
-
-int isprefix(s, pref)
-     char * s;
-     char * pref;
-{
-  while (1) {
-    if (*pref == 0) return 1;
-    if (*s == 0) return 0;
-    if (*s != *pref) return 0;
-    s++;
-    pref++;
-  }
-}
-
-int main(argc, argv)
-     int argc;
-     char * argv [];
-{
-  unsigned char * p;
-  int c;
-  int inquote;
-  int inverb;
-  int inverbatim_like;
-  int incaml;
-  int inverbatim = 0;
-  char *verbatim_end_in = "";
-  char *verbatim_end_out = "";
-
-  for (c = 0; c < 256; c++) transl[c] = NULL;
-#ifdef TIE_BLANKS
-  transl[' '] = "~";
-  transl['\n'] = "~";
-#else
-  transl[' '] = "\\ ";
-  transl['\n'] = "\\ ";
-#endif
-  transl['{'] = "{\\char123}";
-  transl['}'] = "{\\char125}";
-  transl['^'] = "{\\char94}";
-  transl['_'] = "{\\char95}";
-  transl['\\'] = "{\\char92}";
-  transl['~'] = "{\\char126}";
-  transl['$'] = "\\$";
-  transl['&'] = "{\\char38}";
-  transl['#'] = "\\#";
-  transl['%'] = "\\%";
-  transl['\''] = "{\\textquotesingle}";
-  transl['`'] = "{\\textasciigrave}";
-  inverbatim_like = 0;
-  incaml = 0;
-  inquote = 0;
-  inverbatim = 0;
-
-  puts ("% THIS FILE IS GENERATED.\n");
-
-  while(fgets(line, LINE_LENGTH, stdin) != NULL) {
-    if (inverbatim_like) {
-      fputs(line, stdout);
-      if (isprefix(line, "\\end{caml_")
-          || isprefix(line, "\\end{rawhtml}")) inverbatim_like = 0;
-      continue;
-    }
-    if (incaml) {
-      fputs(line, stdout);
-      if (isprefix(line, "\\endcamlexample")) incaml = 0;
-      continue;
-    }
-    if (inverbatim){
-      if (isprefix (line, verbatim_end_in)){
-        fputs (verbatim_end_out, stdout);
-        inverbatim = 0;
-      }else{
-        for (p = (unsigned char *) line; *p != 0; p++){
-          c = *p;
-          if (c == ' ' || c == '\n' || transl[c] == NULL){
-            putchar (c);
-          }else{
-            fputs (transl[c], stdout);
-          }
-        }
-      }
-      continue;
-    }
-    if (isprefix(line, "\\begin{caml_")
-        || isprefix(line, "\\begin{rawhtml}")) {
-      fputs(line, stdout);
-      inverbatim_like = 1;
-      continue;
-    }
-    if (isprefix(line, "\\camlexample")) {
-      fputs(line, stdout);
-      incaml = 1;
-      continue;
-    }
-    if (isprefix (line, "\\begin{verbatim}")){
-      fputs ("\\begin{machineenv}", stdout);
-      inverbatim = 1;
-      verbatim_end_in = "\\end{verbatim}";
-      verbatim_end_out = "\\end{machineenv}";
-      continue;
-    }
-    if (isprefix (line, "\\begin{ocamldoccode}")){
-      fputs ("\\begin{ocamldoccode}", stdout);
-      inverbatim = 1;
-      verbatim_end_in = "\\end{ocamldoccode}";
-      verbatim_end_out = "\\end{ocamldoccode}";
-      continue;
-    }
-    inverb = 0;
-    for (p = (unsigned char *) line; *p != 0; p++) {
-      c = *p;
-      if (inverb) {
-        if (c == inverb){
-          inverb = 0;
-        }else if (c == '\'' || c == '`'){
-          fprintf (stderr, "Warning: %c found in \\verb\n", c);
-        }
-        putchar(c);
-        continue;
-      }
-      switch(c) {
-      case '"':
-        if (inquote) {
-          fputs("}}", stdout);
-          inquote = 0;
-        } else {
-          fputs("{\\machine{", stdout);
-          inquote = 1;
-        }
-        break;
-      case '\\':
-        if (inquote) {
-          if (p[1] == '"' || p[1] == '\\') {
-            c = p[1];
-            p++;
-          }
-          if (transl[c] != NULL)
-            fputs(transl[c], stdout);
-          else
-            putchar(c);
-        } else if (isprefix(p, "\\verb") && p[5] != 0 && !isalpha(p[5])) {
-          inverb = p[5];
-          p = p + 5;
-          fputs("\\verb", stdout);
-          putchar(inverb);
-        } else {
-          putchar('\\');
-        }
-        break;
-      default:
-        if (inquote && transl[c] != NULL)
-          fputs(transl[c], stdout);
-        else
-          putchar(c);
-      }
-    }
-  }
-  return 0;
-}

manual/tools/texquote2.ml

@@ -0,0 +1,137 @@
+type environment =
+  | Normal
+  | Caml
+  | Verbatim of string * string
+  | Verbatim_like
+
+let in_quotes = ref false
+
+let is_alpha c =
+  ('A' <= c && c <= 'Z') || ('a' <= c && c <= 'z')
+
+let is_prefix prefix str =
+  let length_prefix = String.length prefix in
+  let length_str = String.length str in
+  if length_prefix > length_str
+  then false
+  else (String.sub str 0 length_prefix) = prefix
+
+let escape = function
+  | ' ' | '\n' -> "\\ "
+  | '{' -> "{\\char123}"
+  | '}' -> "{\\char125}"
+  | '^' -> "{\\char94}"
+  | '_' -> "{\\char95}"
+  | '\\' -> "{\\char92}"
+  | '~' -> "{\\char126}"
+  | '$' -> "\\$"
+  | '&' -> "{\\char38}"
+  | '#' -> "\\#"
+  | '%' -> "\\%"
+  | '\'' -> "{\\textquotesingle}"
+  | '`' -> "{\\textasciigrave}"
+  | _ -> ""
+
+let process_normal_line line =
+  let (verb_mark : char option ref) = ref None in
+  let l = String.length line in
+  let i = ref 0 in
+  while !i<l do
+    match !verb_mark with
+    | None ->
+      (match line.[!i] with
+      | '"' ->
+        let r = if !in_quotes then "}}" else "{\\machine{" in
+        print_string r;
+        in_quotes := not !in_quotes;
+        incr i;
+      | '\\' ->
+        if !in_quotes
+        then begin
+          if (!i < l-1) && (line.[!i+1] = '"' || line.[!i+1] = '\\')
+          then incr i;
+          let t = escape line.[!i] in
+          if t<>"" then print_string t else print_char line.[!i];
+          incr i;
+        end else if is_prefix "\\verb" (String.sub line !i (l - !i))
+                    && not (is_alpha line.[!i+5])
+        then begin
+          i := !i+5;
+          verb_mark := Some line.[!i];
+          print_string "\\verb";
+          print_char line.[!i];
+          incr i;
+        end else (print_char '\\'; incr i)
+      | _ ->
+        if !in_quotes && (escape line.[!i] <> "")
+        then print_string (escape line.[!i])
+        else print_char line.[!i];
+        incr i;
+      )
+    | Some mark ->
+      if line.[!i] = mark
+      then verb_mark := None
+      else if line.[!i] = '\'' || line.[!i] = '`'
+      then Printf.eprintf "Warning: %c found in \\verb\n" line.[!i];
+      print_char line.[!i];
+      incr i;
+  done
+
+let process_line line = function
+  | Normal ->
+    if is_prefix "\\begin{caml_" line || is_prefix "\\begin{rawhtml}" line
+    then (print_string line; Verbatim_like)
+    else if is_prefix "\\camlexample" line
+    then (print_endline line; Caml)
+    else if is_prefix "\\begin{verbatim}" line
+    then begin
+      print_string "\\begin{machineenv}";
+      (Verbatim ("\\end{verbatim}", "\\end{machineenv}"))
+    end else if is_prefix "\\begin{ocamldoccode}" line
+    then begin
+      print_string "\\begin{ocamldoccode}";
+      (Verbatim ("\\end{ocamldoccode}", "\\end{ocamldoccode}"))
+    end else begin
+      process_normal_line line;
+      if !in_quotes
+      then print_string (escape '\n')
+      else print_newline();
+      Normal
+    end
+  | Caml ->
+    print_endline line;
+    if is_prefix "\\endcamlexample" line then Normal else Caml
+  | Verbatim (verbatim_end_in, verbatim_end_out) as env ->
+    if is_prefix verbatim_end_in line
+    then begin
+      print_string verbatim_end_out;
+      Normal
+    end else begin
+      for i=0 to (String.length line) - 1 do
+        let c = line.[i] in
+        let t = escape c in
+        if c=' ' || c='\n' || t=""
+        then print_char c
+        else print_string t
+      done;
+      print_newline();
+      env
+    end
+  | Verbatim_like ->
+    print_endline line;
+    if is_prefix "\\end{caml_" line || is_prefix "\\end{rawhtml}" line
+    then Normal
+    else Verbatim_like
+
+let rec process_input env = match input_line stdin with
+  | exception End_of_file -> ()
+  | line ->
+    let env = process_line line env in
+    process_input env
+
+let main() =
+  print_endline "% THIS FILE IS GENERATED.";
+  print_newline();
+  process_input Normal
+
+let _ = main()