First backend! Text output now pretty much works.

[originally from svn r240]

12 files changed, 643 insertions, 51 deletions

diff --git a/Makefile b/Makefile
index d89b105..89843cc 100644
--- a/Makefile
+++ b/Makefile
@@ -45,6 +45,7 @@ SRC := ../
 
 MODULES := main malloc ustring error help licence version misc tree23
 MODULES += input keywords contents index style biblio
+MODULES += bk_text
 
 OBJECTS := $(addsuffix .o,$(MODULES))
 DEPS := $(addsuffix .d,$(MODULES))
diff --git a/biblio.c b/biblio.c
index 70e9411..ffff435 100644
--- a/biblio.c
+++ b/biblio.c
@@ -99,6 +99,7 @@ void gen_citations(paragraph *source, keywordlist *kl) {
 		wd->next = NULL;
 		kw->text = wd;
 	    }
+	    para->kwtext = kw->text;
 	}
     }
 }
diff --git a/bk_text.c b/bk_text.c
new file mode 100644
index 0000000..786c126
--- /dev/null
+++ b/bk_text.c
@@ -0,0 +1,444 @@
+/*
+ * text backend for Buttress
+ */
+
+#include <stdio.h>
+#include <stdlib.h>
+#include "buttress.h"
+
+typedef enum { LEFT, CENTRE } alignment;
+
+typedef struct {
+    int indent;
+    int listindentbefore, listindentafter;
+    int width;
+    alignment titlealign, chapteralign;
+    wchar_t titleunderline, chapterunderline;
+    int include_version_id;
+    int indent_preambles;
+    word bullet;
+} textconfig;
+
+static int text_convert(wchar_t *, char **);
+
+static void text_title(FILE *, word *, word *, alignment, wchar_t, int, int);
+static void text_heading(FILE *, word *, word *, int, int);
+static void text_para(FILE *, word *, char *, word *, int, int, int);
+static void text_codepara(FILE *, word *, int, int);
+static void text_versionid(FILE *, word *);
+
+static textconfig text_configure(paragraph *sourceform) {
+    textconfig ret;
+
+    /*
+     * Non-negotiables.
+     */
+    ret.bullet.next = NULL;
+    ret.bullet.alt = NULL;
+    ret.bullet.type = word_Normal;
+
+    /*
+     * Defaults.
+     */
+    ret.indent = 7;
+    ret.listindentbefore = 1;
+    ret.listindentafter = 3;
+    ret.width = 68;
+    ret.titlealign = CENTRE;
+    ret.titleunderline = L'=';
+    ret.chapteralign = LEFT;
+    ret.chapterunderline = L'-';
+    ret.include_version_id = TRUE;
+    ret.indent_preambles = FALSE;
+    ret.bullet.text = ustrdup(L"-");
+
+    /*
+     * FIXME: must walk the source form gleaning configuration from
+     * \config paragraphs.
+     */
+    IGNORE(sourceform);		       /* for now */
+
+    return ret;
+}
+
+void text_backend(paragraph *sourceform, keywordlist *keywords, index *idx) {
+    paragraph *p;
+    textconfig conf;
+    word *prefix, *body, *wp;
+    word spaceword;
+    FILE *fp;
+    char *prefixextra;
+    int indentb, indenta;
+
+    IGNORE(keywords);		       /* we don't happen to need this */
+    IGNORE(idx);		       /* or this */
+
+    conf = text_configure(sourceform);
+
+    /*
+     * Determine the output file name, and open the output file
+     *
+     * FIXME: want configurable output file names here. For the
+     * moment, we'll just call it `output.txt'.
+     */
+    fp = fopen("output.txt", "w");
+    if (!fp) {
+	error(err_cantopenw, "output.txt");
+	return;
+    }
+
+    /* Do the title */
+    for (p = sourceform; p; p = p->next)
+	if (p->type == para_Title)
+	    text_title(fp, NULL, p->words,
+		       conf.titlealign, conf.titleunderline,
+		       conf.indent, conf.width);
+
+    /* Do the preamble and copyright */
+    for (p = sourceform; p; p = p->next)
+	if (p->type == para_Preamble)
+	    text_para(fp, NULL, NULL, p->words,
+		      conf.indent_preambles ? conf.indent : 0, 0,
+		      conf.width + (conf.indent_preambles ? 0 : conf.indent));
+    for (p = sourceform; p; p = p->next)
+	if (p->type == para_Copyright)
+	    text_para(fp, NULL, NULL, p->words,
+		      conf.indent_preambles ? conf.indent : 0, 0,
+		      conf.width + (conf.indent_preambles ? 0 : conf.indent));
+
+    /* Do the main document */
+    for (p = sourceform; p; p = p->next) switch (p->type) {
+
+	/*
+	 * Things we ignore because we've already processed them or
+	 * aren't going to touch them in this pass.
+	 */
+      case para_IM:
+      case para_BR:
+      case para_Biblio:		       /* only touch BiblioCited */
+      case para_VersionID:
+      case para_Copyright:
+      case para_Preamble:
+      case para_NoCite:
+      case para_Title:
+	break;
+
+	/*
+	 * Chapter titles.
+	 */
+      case para_Chapter:
+      case para_Appendix:
+      case para_UnnumberedChapter:
+	text_title(fp, p->kwtext, p->words,
+		   conf.chapteralign, conf.chapterunderline,
+		   conf.indent, conf.width);
+	break;
+
+      case para_Heading:
+      case para_Subsect:
+	text_heading(fp, p->kwtext2, p->words, conf.indent, conf.width);
+	break;
+
+      case para_Normal:
+      case para_BiblioCited:	       /* FIXME: put the citation on front */
+      case para_Bullet:
+      case para_NumberedList:
+	if (p->type == para_Bullet) {
+	    prefix = &conf.bullet;
+	    prefixextra = NULL;
+	    indentb = conf.listindentbefore;
+	    indenta = conf.listindentafter;
+	} else if (p->type == para_NumberedList) {
+	    prefix = p->kwtext;
+	    prefixextra = ".";
+	    indentb = conf.listindentbefore;
+	    indenta = conf.listindentafter;
+	} else {
+	    prefix = NULL;
+	    prefixextra = NULL;
+	    indentb = indenta = 0;
+	}
+	if (p->type == para_BiblioCited) {
+	    body = dup_word_list(p->kwtext);
+	    for (wp = body; wp->next; wp = wp->next);
+	    wp->next = &spaceword;
+	    spaceword.next = p->words;
+	    spaceword.alt = NULL;
+	    spaceword.type = word_WhiteSpace;
+	    spaceword.text = NULL;
+	} else {
+	    wp = NULL;
+	    body = p->words;
+	}
+	text_para(fp, prefix, prefixextra, body,
+		  conf.indent + indentb, indenta, conf.width);
+	if (wp) {
+	    wp->next = NULL;
+	    free_word_list(body);
+	}
+	break;
+
+      case para_Code:
+	text_codepara(fp, p->words, conf.indent, conf.width);
+	break;
+    }
+
+    /* Do the version ID */
+    if (conf.include_version_id) {
+	for (p = sourceform; p; p = p->next)
+	    if (p->type == para_VersionID)
+ 		text_versionid(fp, p->words);
+    }
+
+    /*
+     * Tidy up
+     */
+    fclose(fp);
+    sfree(conf.bullet.text);
+}
+
+/*
+ * Convert a wide string into a string of chars. If `result' is
+ * non-NULL, mallocs the resulting string and stores a pointer to
+ * it in `*result'. If `result' is NULL, merely checks whether all
+ * characters in the string are feasible for the output character
+ * set.
+ *
+ * Return is nonzero if all characters are OK. If not all
+ * characters are OK but `result' is non-NULL, a result _will_
+ * still be generated!
+ */
+static int text_convert(wchar_t *s, char **result) {
+    /*
+     * FIXME. Currently this is ISO8859-1 only.
+     */
+    int doing = (result != 0);
+    int ok = TRUE;
+    char *p = NULL;
+    int plen = 0, psize = 0;
+
+    for (; *s; s++) {
+	wchar_t c = *s;
+	char outc;
+
+	if ((c >= 32 && c <= 126) ||
+	    (c >= 160 && c <= 255)) {
+	    /* Char is OK. */
+	    outc = (char)c;
+	} else {
+	    /* Char is not OK. */
+	    ok = FALSE;
+	    outc = 0xBF;	       /* approximate the good old DEC `uh?' */
+	}
+	if (doing) {
+	    if (plen >= psize) {
+		psize = plen + 256;
+		p = resize(p, psize);
+	    }
+	    p[plen++] = outc;
+	}
+    }
+    if (doing) {
+	p = resize(p, plen+1);
+	p[plen] = '\0';
+	*result = p;
+    }
+    return ok;
+}
+
+static void text_rdaddwc(rdstringc *rs, word *text, word *end) {
+    char *c;
+
+    for (; text && text != end; text = text->next) switch (text->type) {
+      case word_HyperLink:
+      case word_HyperEnd:
+      case word_UpperXref:
+      case word_LowerXref:
+      case word_XrefEnd:
+      case word_IndexRef:
+	break;
+
+      case word_Normal:
+      case word_Emph:
+      case word_Code:
+      case word_WeakCode:
+	if (text->type == word_Emph)
+	    rdaddc(rs, '_');	       /* FIXME: configurability */
+	else if (text->type == word_Code)
+	    rdaddc(rs, '`');	       /* FIXME: configurability */
+	if (text_convert(text->text, &c))
+	    rdaddsc(rs, c);
+	else
+	    text_rdaddwc(rs, text->alt, NULL);
+	sfree(c);
+	if (text->type == word_Emph)
+	    rdaddc(rs, '_');	       /* FIXME: configurability */
+	else if (text->type == word_Code)
+	    rdaddc(rs, '\'');	       /* FIXME: configurability */
+	break;
+
+      case word_WhiteSpace:
+	rdaddc(rs, ' ');
+	break;
+    }
+}
+
+static int text_width(word *);
+
+static int text_width_list(word *text) {
+    int w = 0;
+    while (text) {
+	w += text_width(text);
+	text = text->next;
+    }
+    return w;
+}
+
+static int text_width(word *text) {
+    switch (text->type) {
+      case word_HyperLink:
+      case word_HyperEnd:
+      case word_UpperXref:
+      case word_LowerXref:
+      case word_XrefEnd:
+      case word_IndexRef:
+	return 0;
+
+      case word_Normal:
+      case word_Emph:
+      case word_Code:
+      case word_WeakCode:
+	return ((text->type == word_Emph || text->type == word_Code ? 2 : 0) +
+		(text_convert(text->text, NULL) ?
+		 ustrlen(text->text) :
+		 text_width_list(text->alt)));
+
+      case word_WhiteSpace:
+	return 1;
+    }
+    return 0;			       /* should never happen */
+}
+
+static void text_title(FILE *fp, word *prefix, word *text,
+		       alignment align, wchar_t underline,
+		       int indent, int width) {
+    rdstringc t = { 0, 0, NULL };
+    int margin, length;
+
+    if (prefix) {
+	text_rdaddwc(&t, prefix, NULL);
+	rdaddsc(&t, ": ");
+    }
+    text_rdaddwc(&t, text, NULL);
+
+    length = strlen(t.text);
+    if (align == CENTRE) {
+	margin = (indent + width - length)/2;
+	if (margin < 0) margin = 0;
+    } else
+	margin = 0;
+
+    fprintf(fp, "%*s%s\n", margin, "", t.text);
+    if (underline != L'\0') {
+	char *u, uc;
+	wchar_t uw[2];
+	uw[0] = underline; uw[1] = L'\0';
+	text_convert(uw, &u);
+	uc = u[0];
+	sfree(u);
+	fprintf(fp, "%*s", margin, "");
+	while (length--)
+	    putc(uc, fp);
+	putc('\n', fp);
+    }
+    putc('\n', fp);
+
+    sfree(t.text);
+}
+
+static void text_heading(FILE *fp, word *prefix, word *text,
+			 int indent, int width) {
+    rdstringc t = { 0, 0, NULL };
+    int margin;
+
+    if (prefix) {
+	text_rdaddwc(&t, prefix, NULL);
+	rdaddc(&t, ' ');
+	margin = strlen(t.text);
+    }
+    text_rdaddwc(&t, text, NULL);
+
+    margin = indent - margin;
+    if (margin < 0) margin = 0;
+
+    fprintf(fp, "%*s%s\n\n", margin, "", t.text);
+
+    if (strlen(t.text) > (size_t)width) {
+	/* FIXME: warn */
+    }
+
+    sfree(t.text);
+}
+
+static void text_para(FILE *fp, word *prefix, char *prefixextra, word *text,
+		      int indent, int extraindent, int width) {
+    wrappedline *wrapping, *p;
+    rdstringc pfx = { 0, 0, NULL };
+    int e;
+    int firstlinewidth = width;
+
+    if (prefix) {
+	text_rdaddwc(&pfx, prefix, NULL);
+	if (prefixextra)
+	    rdaddsc(&pfx, prefixextra);
+	fprintf(fp, "%*s%s", indent, "", pfx.text);
+	e = extraindent - strlen(pfx.text);
+	if (e < 0) {
+	    e = 0;
+	    firstlinewidth -= e;
+	    if (firstlinewidth < 0) {
+		e = indent + extraindent;
+		firstlinewidth = width;
+		fprintf(fp, "\n");
+	    }
+	}
+	sfree(pfx.text);
+    } else
+	e = indent + extraindent;
+
+    wrapping = wrap_para(text, firstlinewidth, width, text_width);
+    for (p = wrapping; p; p = p->next) {
+	rdstringc t = { 0, 0, NULL };
+	text_rdaddwc(&t, p->begin, p->end);
+	fprintf(fp, "%*s%s\n", e, "", t.text);
+	e = indent + extraindent;
+	sfree(t.text);
+    }
+    wrap_free(wrapping);
+    putc('\n', fp);
+}
+
+static void text_codepara(FILE *fp, word *text, int indent, int width) {
+    for (; text; text = text->next) if (text->type == word_WeakCode) {
+	char *c;
+	text_convert(text->text, &c);
+	if (strlen(c) > (size_t)width) {
+	    /* FIXME: warn */
+	}
+	fprintf(fp, "%*s%s\n", indent, "", c);
+	sfree(c);
+    }
+
+    putc('\n', fp);
+}
+
+static void text_versionid(FILE *fp, word *text) {
+    rdstringc t = { 0, 0, NULL };
+
+    rdaddc(&t, '[');		       /* FIXME: configurability */
+    text_rdaddwc(&t, text, NULL);
+    rdaddc(&t, ']');		       /* FIXME: configurability */
+
+    fprintf(fp, "%s\n\n", t.text);
+    sfree(t.text);
+}
diff --git a/buttress.h b/buttress.h
index 9bdfcfc..062bc37 100644
--- a/buttress.h
+++ b/buttress.h
@@ -16,6 +16,9 @@
 #define FALSE 0
 #endif
 
+/* For suppressing unused-parameter warnings */
+#define IGNORE(x) ( (x) = (x) )
+
 /*
  * Structure tags
  */
@@ -65,6 +68,7 @@ struct paragraph_Tag {
     word *words;		       /* list of words in paragraph */
     int aux;			       /* number, in a numbered paragraph */
     word *kwtext;		       /* chapter/section indication */
+    word *kwtext2;		       /* numeric-only form of kwtext */
     filepos fpos;
 };
 enum {
@@ -140,7 +144,8 @@ enum {
     err_nestedindex,		       /* unable to nest `\i' thingys */
     err_nosuchkw,		       /* unresolved cross-reference */
     err_multiBR,		       /* multiple \BRs on same keyword */
-    err_nosuchidxtag		       /* \IM on unknown index tag (warning) */
+    err_nosuchidxtag,		       /* \IM on unknown index tag (warning) */
+    err_cantopenw		       /* can't open output file for write */
 };
 
 /*
@@ -205,8 +210,36 @@ stack stk_new(void);
 void stk_free(stack);
 void stk_push(stack, void *);
 void *stk_pop(stack);
+
+typedef struct tagRdstring rdstring;
+struct tagRdstring {
+    int pos, size;
+    wchar_t *text;
+};
+typedef struct tagRdstringc rdstringc;
+struct tagRdstringc {
+    int pos, size;
+    char *text;
+};
+void rdadd(rdstring *rs, wchar_t c);
+void rdadds(rdstring *rs, wchar_t *p);
+wchar_t *rdtrim(rdstring *rs);
+void rdaddc(rdstringc *rs, char c);
+void rdaddsc(rdstringc *rs, char *p);
+char *rdtrimc(rdstringc *rs);
+
 int compare_wordlists(word *a, word *b);
 
+typedef struct tagWrappedLine wrappedline;
+struct tagWrappedLine {
+    wrappedline *next;
+    word *begin, *end;		       /* first & last words of line */
+    int nspaces;		       /* number of whitespaces in line */
+    int shortfall;		       /* how much shorter than max width */
+};
+wrappedline *wrap_para(word *, int, int, int (*)(word *));
+void wrap_free(wrappedline *);
+
 /*
  * input.c
  */
@@ -243,12 +276,13 @@ void cleanup_index(index *);
  * takes responsibility for arg 2 */
 void index_merge(index *, int is_explicit, wchar_t *, word *);
 void build_index(index *);
+void index_debug(index *);
 
 /*
  * contents.c
  */
 numberstate *number_init(void);
-word *number_mktext(numberstate *, int, int, int);
+word *number_mktext(numberstate *, int, int, int, word **);
 void number_free(numberstate *);
 
 /*
@@ -275,6 +309,11 @@ void freetree23(tree23 *);
 void *add23(tree23 *, void *, int (*cmp)(void *, void *));
 void *find23(tree23 *, void *, int (*cmp)(void *, void *));
 void *first23(tree23 *, enum23 *);
-void *next23(tree23 *, enum23 *);
+void *next23(enum23 *);
+
+/*
+ * bk_text.c
+ */
+void text_backend(paragraph *, keywordlist *, index *);
 
 #endif
diff --git a/contents.c b/contents.c
index 790c58f..494e987 100644
--- a/contents.c
+++ b/contents.c
@@ -89,8 +89,10 @@ static void doanumber(word ***wret, int num) {
     dotext(wret, p);
 }
 
-word *number_mktext(numberstate *state, int para, int aux, int prev) {
+word *number_mktext(numberstate *state, int para, int aux, int prev,
+		    word **auxret) {
     word *ret = NULL;
+    word **ret2 = &ret;
     word **pret = &ret;
     int i, level;
 
@@ -101,6 +103,7 @@ word *number_mktext(numberstate *state, int para, int aux, int prev) {
 	    state->sectionlevels[i] = 0;
 	dotext(&pret, L"Chapter");
 	dospace(&pret);
+	ret2 = pret;
 	donumber(&pret, state->chapternum);
 	state->ischapter = 1;
 	break;
@@ -117,6 +120,7 @@ word *number_mktext(numberstate *state, int para, int aux, int prev) {
 	    state->sectionlevels[i] = 0;
 	dotext(&pret, L"Section");
 	dospace(&pret);
+	ret2 = pret;
 	if (state->ischapter)
 	    donumber(&pret, state->chapternum);
 	else
@@ -134,10 +138,12 @@ word *number_mktext(numberstate *state, int para, int aux, int prev) {
 	    state->sectionlevels[i] = 0;
 	dotext(&pret, L"Appendix");
 	dospace(&pret);
+	ret2 = pret;
 	doanumber(&pret, state->appendixnum);
 	state->ischapter = 0;
 	break;
       case para_NumberedList:
+	ret2 = pret;
 	if (prev != para_NumberedList)
 	    state->listitem = 0;
 	state->listitem++;
@@ -145,5 +151,7 @@ word *number_mktext(numberstate *state, int para, int aux, int prev) {
 	break;
     }
 
+    if (auxret)
+	*auxret = *ret2;
     return ret;
 }
diff --git a/error.c b/error.c
index 704ae03..b7eb0d5 100644
--- a/error.c
+++ b/error.c
@@ -142,6 +142,11 @@ static void do_error(int code, va_list ap) {
 	flags = 0;
 	/* FIXME: need to get a filepos to here somehow */
 	break;
+      case err_cantopenw:
+	sp = va_arg(ap, char *);
+	sprintf(error, "unable to open output file `%.200s'", sp);
+	flags = PREFIX;
+	break;
     }
 
     if (flags & PREFIX)
diff --git a/index.c b/index.c
index 3f629a0..4d7cfe9 100644
--- a/index.c
+++ b/index.c
@@ -128,12 +128,13 @@ void index_merge(index *idx, int is_explicit, wchar_t *tags, word *text) {
  * entries.
  */
 void build_index(index *i) {
-    indextag *t, **ta;
+    indextag *t;
+    word **ta;
     enum23 e;
     int j;
 
     for (t = (indextag *)first23(i->tags, &e); t;
-	 t = (indextag *)next23(i->tags, &e)) {
+	 t = (indextag *)next23(&e)) {
 	if (t->implicit_text) {
 	    t->nrefs = 1;
 	    ta = &t->implicit_text;
@@ -156,10 +157,9 @@ void cleanup_index(index *i) {
     indextag *t;
     indexentry *ent;
     enum23 e;
-    int j;
 
     for (t = (indextag *)first23(i->tags, &e); t;
-	 t = (indextag *)next23(i->tags, &e)) {
+	 t = (indextag *)next23(&e)) {
 	sfree(t->name);
 	free_word_list(t->implicit_text);
 	sfree(t->explicit_texts);
@@ -168,30 +168,29 @@ void cleanup_index(index *i) {
     }
     freetree23(i->tags);
     for (ent = (indexentry *)first23(i->entries, &e); ent;
-	 ent = (indexentry *)next23(i->entries, &e)) {
+	 ent = (indexentry *)next23(&e)) {
 	sfree(ent);
     }
     freetree23(i->entries);
     sfree(i);
 }
 
+static void dbg_prtwordlist(int level, word *w);
+static void dbg_prtmerge(int is_explicit, wchar_t *tag, word *text);
+
 void index_debug(index *i) {
     indextag *t;
     enum23 e;
     int j;
 
     for (t = (indextag *)first23(i->tags, &e); t;
-	 t = (indextag *)next23(i->tags, &e)) {
+	 t = (indextag *)next23(&e)) {
 	if (t->implicit_text)
 	    dbg_prtmerge(0, t->name, t->implicit_text);
 	for (j = 0; j < t->nexplicit; j++)
 	    dbg_prtmerge(1, t->name, t->explicit_texts[j]);
     }
 }
-#define DEBUG
-
-#ifdef DEBUG
-static void dbg_prtwordlist(int level, word *w);
 
 static void dbg_prtmerge(int is_explicit, wchar_t *tag, word *text) {
     printf("\\IM: %splicit: \"", is_explicit ? "ex" : "im");
@@ -221,4 +220,3 @@ static void dbg_prtwordlist(int level, word *w) {
 	printf("\n");
     }
 }
-#endif
diff --git a/input.c b/input.c
index c5d0b5c..d5c4685 100644
--- a/input.c
+++ b/input.c
@@ -76,36 +76,6 @@ static int get(input *in) {
 }
 
 /*
- * Small routines to amalgamate a string from an input source.
- */
-typedef struct tagRdstring rdstring;
-struct tagRdstring {
-    int pos, size;
-    wchar_t *text;
-};
-static void rdadd(rdstring *rs, wchar_t c) {
-    if (rs->pos >= rs->size-1) {
-	rs->size = rs->pos + 128;
-	rs->text = resize(rs->text, rs->size);
-    }
-    rs->text[rs->pos++] = c;
-    rs->text[rs->pos] = 0;
-}
-static void rdadds(rdstring *rs, wchar_t *p) {
-    int len = ustrlen(p);
-    if (rs->pos >= rs->size - len) {
-	rs->size = rs->pos + len + 128;
-	rs->text = resize(rs->text, rs->size);
-    }
-    ustrcpy(rs->text + rs->pos, p);
-    rs->pos += len;
-}
-static wchar_t *rdtrim(rdstring *rs) {
-    rs->text = resize(rs->text, rs->pos + 1);
-    return rs->text;
-}
-
-/*
  * Lexical analysis of source files.
  */
 typedef struct token_Tag token;
diff --git a/keywords.c b/keywords.c
index c446761..9019b55 100644
--- a/keywords.c
+++ b/keywords.c
@@ -100,7 +100,7 @@ keywordlist *get_keywords(paragraph *source) {
 	 * Number the chapter / section / list-item / whatever.
 	 */
 	source->kwtext = number_mktext(n, source->type, source->aux,
-				       prevpara);
+				       prevpara, &source->kwtext2);
 	prevpara = source->type;
 
 	if (source->keyword && *source->keyword) {
@@ -161,7 +161,9 @@ void subst_keywords(paragraph *source, keywordlist *kl) {
 		} else
 		    subst = dup_word_list(kw->text);
 
-		if (subst && ptr->type == word_LowerXref)
+		if (subst && ptr->type == word_LowerXref &&
+		    kw->para->type != para_Biblio &&
+		    kw->para->type != para_BiblioCited)
 		    ustrlow(subst->text);
 
 		close = mknew(word);
diff --git a/main.c b/main.c
index 1fef5e4..34ce7d4 100644
--- a/main.c
+++ b/main.c
@@ -17,6 +17,7 @@ int main(int argc, char **argv) {
     int nogo;
     int errs;
     int reportcols;
+    int debug;
 
     /*
      * Set up initial (default) parameters.
@@ -26,6 +27,7 @@ int main(int argc, char **argv) {
     nfiles = 0;
     nogo = errs = FALSE;
     reportcols = 0;
+    debug = 0;
 
     if (argc == 1) {
 	usage();
@@ -85,6 +87,7 @@ int main(int argc, char **argv) {
 		  case 'V':
 		  case 'L':
 		  case 'P':
+		  case 'd':
 		    /*
 		     * Option requiring no parameter.
 		     */
@@ -104,6 +107,9 @@ int main(int argc, char **argv) {
 		      case 'P':
 			reportcols = 1;
 			break;
+		      case 'd':
+			debug = TRUE;
+			break;
 		    }
 		    break;
 		  case 'o':
@@ -193,10 +199,14 @@ int main(int argc, char **argv) {
 		index_merge(idx, TRUE, p->keyword, p->words);
 
 	build_index(idx);
-	index_debug(idx);
 
-	dbg_prtkws(keywords);
-	dbg_prtsource(sourceform);
+	if (debug) {
+	    index_debug(idx);
+	    dbg_prtkws(keywords);
+	    dbg_prtsource(sourceform);
+	}
+
+	text_backend(sourceform, keywords, idx);
 
 	free_para_list(sourceform);
 	free_keywords(keywords);
diff --git a/misc.c b/misc.c
index 3ead01c..3d512c6 100644
--- a/misc.c
+++ b/misc.c
@@ -41,6 +41,53 @@ void *stk_pop(stack s) {
 	return NULL;
 }
 
+/*
+ * Small routines to amalgamate a string from an input source.
+ */
+void rdadd(rdstring *rs, wchar_t c) {
+    if (rs->pos >= rs->size-1) {
+	rs->size = rs->pos + 128;
+	rs->text = resize(rs->text, rs->size);
+    }
+    rs->text[rs->pos++] = c;
+    rs->text[rs->pos] = 0;
+}
+void rdadds(rdstring *rs, wchar_t *p) {
+    int len = ustrlen(p);
+    if (rs->pos >= rs->size - len) {
+	rs->size = rs->pos + len + 128;
+	rs->text = resize(rs->text, rs->size);
+    }
+    ustrcpy(rs->text + rs->pos, p);
+    rs->pos += len;
+}
+wchar_t *rdtrim(rdstring *rs) {
+    rs->text = resize(rs->text, rs->pos + 1);
+    return rs->text;
+}
+
+void rdaddc(rdstringc *rs, char c) {
+    if (rs->pos >= rs->size-1) {
+	rs->size = rs->pos + 128;
+	rs->text = resize(rs->text, rs->size);
+    }
+    rs->text[rs->pos++] = c;
+    rs->text[rs->pos] = 0;
+}
+void rdaddsc(rdstringc *rs, char *p) {
+    int len = strlen(p);
+    if (rs->pos >= rs->size - len) {
+	rs->size = rs->pos + len + 128;
+	rs->text = resize(rs->text, rs->size);
+    }
+    strcpy(rs->text + rs->pos, p);
+    rs->pos += len;
+}
+char *rdtrimc(rdstringc *rs) {
+    rs->text = resize(rs->text, rs->pos + 1);
+    return rs->text;
+}
+
 int compare_wordlists(word *a, word *b) {
     int t;
     while (a && b) {
@@ -84,3 +131,70 @@ int compare_wordlists(word *a, word *b) {
     else
 	return 0;
 }
+
+wrappedline *wrap_para(word *text, int width, int subsequentwidth,
+		       int (*widthfn)(word *)) {
+    wrappedline *head = NULL, **ptr = &head;
+    word *spc;
+    int nspc;
+    int thiswidth, lastgood;
+
+    while (text) {
+	wrappedline *w = mknew(wrappedline);
+	*ptr = w;
+	ptr = &w->next;
+	w->next = NULL;
+
+	w->begin = text;
+	spc = NULL;
+	nspc = 0;
+	thiswidth = lastgood = 0;
+	while (text) {
+	    thiswidth += widthfn(text);
+	    if (text->next && text->next->type == word_WhiteSpace) {
+		if (thiswidth > width)
+		    break;
+		spc = text->next;
+		lastgood = thiswidth;
+		nspc++;
+	    }
+	    text = text->next;
+	}
+	/*
+	 * We've exited this loop on one of three conditions. spc
+	 * might be non-NULL and we've overflowed: we have broken
+	 * the paragraph there. spc might be NULL and text might be
+	 * NULL too: we've reached the end of the paragraph and
+	 * should output the last line. Or text might be non-NULL
+	 * and spc might be NULL: we've got an anomalously long
+	 * line with no spaces we could have broken at. Output it
+	 * anyway as the best we can do.
+	 */
+	if (spc && thiswidth > width) {
+	    w->end = spc;
+	    text = spc->next;
+	    w->nspaces = nspc-1;
+	    w->shortfall = width - lastgood;
+	} else if (!text) {
+	    w->end = NULL;	       /* no end marker needed */
+	    w->nspaces = nspc;
+	    w->shortfall = width - thiswidth;
+	} else {
+	    w->end = text;
+	    text = text->next;
+	    w->nspaces = 0;
+	    w->shortfall = width - thiswidth;
+	}
+	width = subsequentwidth;
+    }
+
+    return head;
+}
+
+void wrap_free(wrappedline *w) {
+    while (w) {
+	wrappedline *t = w->next;
+	sfree(w);
+	w = t;
+    }
+}
diff --git a/tree23.c b/tree23.c
index 4a0d60c..7089c29 100644
--- a/tree23.c
+++ b/tree23.c
@@ -238,7 +238,7 @@ void *first23(tree23 *t, enum23 *e) {
     return n->elems[0];
 }
 
-void *next23(tree23 *t, enum23 *e) {
+void *next23(enum23 *e) {
     node23 *n = (node23 *)e->node;
     int pos = e->posn;