X-Git-Url: https://git.distorted.org.uk/~mdw/sgt/halibut/blobdiff_plain/7136a6c7f094fa423c48ec319748c4fd7e1fa645..refs/heads/master:/main.c

diff --git a/main.c b/main.c
index 531d122..405e6ef 100644
--- a/main.c
+++ b/main.c
@@ -2,6 +2,8 @@
  * main.c: command line parsing and top level
  */
 
+#include <assert.h>
+#include <locale.h>
 #include <stdio.h>
 #include <stdlib.h>
 #include "halibut.h"
@@ -10,24 +12,68 @@ static void dbg_prtsource(paragraph *sourceform);
 static void dbg_prtwordlist(int level, word *w);
 static void dbg_prtkws(keywordlist *kws);
 
+static const struct pre_backend {
+    void *(*func)(paragraph *, keywordlist *, indexdata *);
+    int bitfield;
+} pre_backends[] = {
+    {paper_pre_backend, 0x0001}
+};
+
+static const struct backend {
+    char *name;
+    void (*func)(paragraph *, keywordlist *, indexdata *, void *);
+    paragraph *(*filename)(char *filename);
+    int bitfield, prebackend_bitfield;
+} backends[] = {
+    {"text", text_backend, text_config_filename, 0x0001, 0},
+    {"xhtml", html_backend, html_config_filename, 0x0002, 0},
+    {"html", html_backend, html_config_filename, 0x0002, 0},
+    {"hlp", whlp_backend, whlp_config_filename, 0x0004, 0},
+    {"whlp", whlp_backend, whlp_config_filename, 0x0004, 0},
+    {"winhelp", whlp_backend, whlp_config_filename, 0x0004, 0},
+    {"man", man_backend, man_config_filename, 0x0008, 0},
+    {"info", info_backend, info_config_filename, 0x0010, 0},
+    {"ps", ps_backend, ps_config_filename, 0x0020, 0x0001},
+    {"pdf", pdf_backend, pdf_config_filename, 0x0040, 0x0001},
+};
+
 int main(int argc, char **argv) {
     char **infiles;
-    char *outfile;
     int nfiles;
     int nogo;
     int errs;
     int reportcols;
+    int list_fonts;
+    int input_charset;
     int debug;
+    int backendbits, prebackbits;
+    int k, b;
+    paragraph *cfg, *cfg_tail;
+    void *pre_backend_data[16];
+
+    /*
+     * Use the specified locale everywhere. It'll be used for
+     * output of error messages, and as the default character set
+     * for input files if one is not explicitly specified.
+     * 
+     * However, we need to use standard numeric formatting for
+     * output of things like PDF.
+     */
+    setlocale(LC_ALL, "");
+    setlocale(LC_NUMERIC, "C");
 
     /*
      * Set up initial (default) parameters.
      */
-    infiles = mknewa(char *, argc);
-    outfile = NULL;
+    infiles = snewn(argc, char *);
     nfiles = 0;
     nogo = errs = FALSE;
     reportcols = 0;
+    list_fonts = 0;
+    input_charset = CS_ASCII;
     debug = 0;
+    backendbits = 0;
+    cfg = cfg_tail = NULL;
 
     if (argc == 1) {
 	usage();
@@ -39,7 +85,7 @@ int main(int argc, char **argv) {
      */
     while (--argc) {
 	char *p = *++argv;
-	if (*p == '-') {
+	if (*p == '-' && p[1]) {
 	    /*
 	     * An option.
 	     */
@@ -60,7 +106,38 @@ int main(int argc, char **argv) {
 			    val = p;
 			} else
 			    val = NULL;
-			if (!strcmp(opt, "-help")) {
+
+			assert(opt[0] == '-');
+			for (k = 0; k < (int)lenof(backends); k++)
+			    if (!strcmp(opt+1, backends[k].name)) {
+				backendbits |= backends[k].bitfield;
+				if (val) {
+				    paragraph *p = backends[k].filename(val);
+				    assert(p);
+				    if (cfg_tail)
+					cfg_tail->next = p;
+				    else
+					cfg = p;
+				    while (p->next)
+					p = p->next;
+				    cfg_tail = p;
+				}
+				break;
+			    }
+			if (k < (int)lenof(backends)) {
+			    /* do nothing */;
+			} else if (!strcmp(opt, "-input-charset")) {
+			    if (!val) {
+				errs = TRUE, err_optnoarg(opt);
+			    } else {
+				int charset = charset_from_localenc(val);
+				if (charset == CS_NONE) {
+				    errs = TRUE, err_cmdcharset(val);
+				} else {
+				    input_charset = charset;
+				}
+			    }
+			} else if (!strcmp(opt, "-help")) {
 			    help();
 			    nogo = TRUE;
 			} else if (!strcmp(opt, "-version")) {
@@ -70,15 +147,15 @@ int main(int argc, char **argv) {
 				   !strcmp(opt, "-license")) {
 			    licence();
 			    nogo = TRUE;
-			} else if (!strcmp(opt, "-output")) {
-			    if (!val)
-				errs = TRUE, error(err_optnoarg, opt);
-			    else
-				outfile = val;
+			} else if (!strcmp(opt, "-list-charsets")) {
+			    listcharsets();
+			    nogo = TRUE;
+			} else if (!strcmp(opt, "-list-fonts")) {
+			    list_fonts = TRUE;
 			} else if (!strcmp(opt, "-precise")) {
 			    reportcols = 1;
 			} else {
-			    errs = TRUE, error(err_nosuchopt, opt);
+			    errs = TRUE, err_nosuchopt(opt);
 			}
 		    }
 		    p = NULL;
@@ -112,7 +189,7 @@ int main(int argc, char **argv) {
 			break;
 		    }
 		    break;
-		  case 'o':
+		  case 'C':
 		    /*
 		     * Option requiring parameter.
 		     */
@@ -123,14 +200,50 @@ int main(int argc, char **argv) {
 			char opt[2];
 			opt[0] = c;
 			opt[1] = '\0';
-			errs = TRUE, error(err_optnoarg, opt);
+			errs = TRUE, err_optnoarg(opt);
 		    }
 		    /*
 		     * Now c is the option and p is the parameter.
 		     */
 		    switch (c) {
-		      case 'o':
-			outfile = p;
+		      case 'C':
+			/*
+			 * -C means we split our argument up into
+			 * colon-separated chunks and assemble them
+			 * into a config paragraph.
+			 */
+			{
+			    char *s = dupstr(p), *q, *r;
+			    paragraph *para;
+
+			    para = cmdline_cfg_new();
+
+			    q = r = s;
+			    while (*q) {
+				if (*q == ':') {
+				    *r = '\0';
+				    /* XXX ad-hoc diagnostic */
+				    if (!strcmp(s, "input-charset"))
+					err_futileopt("Cinput-charset",
+					      "; use --input-charset");
+				    cmdline_cfg_add(para, s);
+				    r = s;
+				} else {
+				    if (*q == '\\' && q[1])
+					q++;
+				    *r++ = *q;
+				}
+				q++;
+			    }
+			    *r = '\0';
+			    cmdline_cfg_add(para, s);
+
+			    if (cfg_tail)
+				cfg_tail->next = para;
+			    else
+				cfg = para;
+			    cfg_tail = para;
+			}
 			break;
 		    }
 		    p = NULL;	       /* prevent continued processing */
@@ -143,7 +256,7 @@ int main(int argc, char **argv) {
 			char opt[2];
 			opt[0] = c;
 			opt[1] = '\0';
-			errs = TRUE, error(err_nosuchopt, opt);
+			errs = TRUE, err_nosuchopt(opt);
 		    }
 		}
 	    }
@@ -151,7 +264,10 @@ int main(int argc, char **argv) {
 	    /*
 	     * A non-option argument.
 	     */
-	    infiles[nfiles++] = p;
+	    if (!strcmp(p, "-"))
+		infiles[nfiles++] = NULL;   /* special case: read stdin */
+	    else
+		infiles[nfiles++] = p;
 	}
     }
 
@@ -163,8 +279,8 @@ int main(int argc, char **argv) {
     /*
      * Do the work.
      */
-    if (nfiles == 0) {
-	error(err_noinput);
+    if (nfiles == 0 && !list_fonts) {
+	err_noinput();
 	usage();
 	exit(EXIT_FAILURE);
     }
@@ -183,16 +299,34 @@ int main(int argc, char **argv) {
 	in.pushback = NULL;
 	in.reportcols = reportcols;
 	in.stack = NULL;
+	in.defcharset = input_charset;
 
 	idx = make_index();
 
 	sourceform = read_input(&in, idx);
+	if (list_fonts) {
+	    listfonts();
+	    exit(EXIT_SUCCESS);
+	}
 	if (!sourceform)
 	    exit(EXIT_FAILURE);
 
-	sfree(in.pushback);
+	/*
+	 * Append the config directives acquired from the command
+	 * line.
+	 */
+	{
+	    paragraph *end;
 
-	mark_attr_ends(sourceform);
+	    end = sourceform;
+	    while (end && end->next)
+		end = end->next;
+	    assert(end);
+
+	    end->next = cfg;
+	}
+
+	sfree(in.pushback);
 
 	sfree(infiles);
 
@@ -204,20 +338,65 @@ int main(int argc, char **argv) {
 
 	for (p = sourceform; p; p = p->next)
 	    if (p->type == para_IM)
-		index_merge(idx, TRUE, p->keyword, p->words);
+		index_merge(idx, TRUE, p->keyword, p->words, &p->fpos);
 
 	build_index(idx);
 
+	/*
+	 * Set up attr_First / attr_Last / attr_Always, in the main
+	 * document and in the index entries.
+	 */
+	for (p = sourceform; p; p = p->next)
+	    mark_attr_ends(p->words);
+	{
+	    int i;
+	    indexentry *entry;
+
+	    for (i = 0; (entry = index234(idx->entries, i)) != NULL; i++)
+		mark_attr_ends(entry->text);
+	}
+
 	if (debug) {
 	    index_debug(idx);
 	    dbg_prtkws(keywords);
 	    dbg_prtsource(sourceform);
 	}
 
-	text_backend(sourceform, keywords, idx);
-	xhtml_backend(sourceform, keywords, idx);
-	whlp_backend(sourceform, keywords, idx);
-	man_backend(sourceform, keywords, idx);
+	/*
+	 * Select and run the pre-backends.
+	 */
+	prebackbits = 0;
+	for (k = 0; k < (int)lenof(backends); k++)
+	    if (backendbits == 0 || (backendbits & backends[k].bitfield))
+		prebackbits |= backends[k].prebackend_bitfield;
+	for (k = 0; k < (int)lenof(pre_backends); k++)
+	    if (prebackbits & pre_backends[k].bitfield) {
+		assert(k < (int)lenof(pre_backend_data));
+		pre_backend_data[k] =
+		    pre_backends[k].func(sourceform, keywords, idx);
+	    }
+
+	/*
+	 * Run the selected set of backends.
+	 */
+	for (k = b = 0; k < (int)lenof(backends); k++)
+	    if (b != backends[k].bitfield) {
+		b = backends[k].bitfield;
+		if (backendbits == 0 || (backendbits & b)) {
+		    void *pbd = NULL;
+		    int pbb = backends[k].prebackend_bitfield;
+		    int m;
+
+		    for (m = 0; m < (int)lenof(pre_backends); m++)
+			if (pbb & pre_backends[m].bitfield) {
+			    assert(m < (int)lenof(pre_backend_data));
+			    pbd = pre_backend_data[m];
+			    break;
+			}
+			    
+		    backends[k].func(sourceform, keywords, idx, pbd);
+		}
+	    }
 
 	free_para_list(sourceform);
 	free_keywords(keywords);
@@ -291,6 +470,8 @@ static void dbg_prtwordlist(int level, word *w) {
 	    printf("\"");
 	} else
 	    printf("(no text)");
+	if (w->breaks)
+	    printf(" [breaks]");
 	if (w->alt) {
 	    printf(" alt = {\n");
 	    dbg_prtwordlist(level+1, w->alt);