=================================================================== RCS file: /cvs/docbook2mdoc/docbook2mdoc.c,v retrieving revision 1.9 retrieving revision 1.10 diff -u -p -r1.9 -r1.10 --- docbook2mdoc/docbook2mdoc.c 2014/03/28 13:16:40 1.9 +++ docbook2mdoc/docbook2mdoc.c 2014/03/29 10:56:21 1.10 @@ -1,4 +1,4 @@ -/* $Id: docbook2mdoc.c,v 1.9 2014/03/28 13:16:40 kristaps Exp $ */ +/* $Id: docbook2mdoc.c,v 1.10 2014/03/29 10:56:21 kristaps Exp $ */ /* * Copyright (c) 2014 Kristaps Dzonsons * @@ -78,6 +78,7 @@ struct parse { char *b; /* nil-terminated buffer for pre-print */ size_t bsz; /* current length of b */ size_t mbsz; /* max bsz allocation */ + int newln; /* output: are we on a fresh line */ }; struct node { @@ -132,6 +133,9 @@ static const struct node nodes[NODE__MAX] = { { "title", 0 }, }; +static void +pnode_print(struct parse *p, struct pnode *pn); + /* * Look up whether "parent" is a valid parent for "node". * This is sucked directly from the DocBook specification: look at the @@ -414,7 +418,7 @@ xml_char(void *arg, const XML_Char *p, int sz) * If we're not processing text right now, then create a text * node for doing so. * However, don't do so unless we have some non-whitespace to - * process! + * process: strip out all leading whitespace to be sure. */ if (NODE_TEXT != ps->node) { for (i = 0; i < sz; i++) @@ -422,6 +426,8 @@ xml_char(void *arg, const XML_Char *p, int sz) break; if (i == sz) return; + p += i; + sz -= i; dat = calloc(1, sizeof(struct pnode)); if (NULL == dat) { perror(NULL); @@ -434,7 +440,6 @@ xml_char(void *arg, const XML_Char *p, int sz) TAILQ_INSERT_TAIL(&ps->cur->childq, dat, child); ps->cur = dat; assert(NULL != ps->root); - } /* Append to current buffer. */ @@ -449,6 +454,16 @@ xml_char(void *arg, const XML_Char *p, int sz) ps->cur->bsz += (size_t)sz; } +static void +pnode_trim(struct pnode *pn) +{ + + assert(NODE_TEXT == pn->node); + for ( ; pn->bsz > 0; pn->bsz--) + if ( ! isspace((int)pn->b[pn->bsz - 1])) + break; +} + /* * Begin an element. * First, look for the element. @@ -476,6 +491,7 @@ xml_elem_start(void *arg, const XML_Char *name, const /* Close out text node, if applicable... */ if (NODE_TEXT == ps->node) { assert(NULL != ps->cur); + pnode_trim(ps->cur); ps->cur = ps->cur->parent; assert(NULL != ps->cur); ps->node = ps->cur->node; @@ -541,6 +557,7 @@ xml_elem_end(void *arg, const XML_Char *name) /* Close out text node, if applicable... */ if (NODE_TEXT == ps->node) { assert(NULL != ps->cur); + pnode_trim(ps->cur); ps->cur = ps->cur->parent; assert(NULL != ps->cur); ps->node = ps->cur->node; @@ -694,21 +711,39 @@ pnode_printmacroline(struct parse *p, struct pnode *pn putchar('\n'); } +static void +pnode_printmopen(struct parse *p) +{ + if (p->newln) { + putchar('.'); + p->newln = 0; + } else + putchar(' '); +} + +static void +pnode_printmclose(struct parse *p, int sv) +{ + + if (sv && ! p->newln) { + putchar('\n'); + p->newln = 1; + } +} + /* - * Start the SYNOPSIS macro, unlinking its [superfluous] title. + * If the SYNOPSIS macro has a superfluous title, kill it. */ static void pnode_printrefsynopsisdiv(struct parse *p, struct pnode *pn) { struct pnode *pp; - TAILQ_FOREACH(pp, &pn->childq, child) + TAILQ_FOREACH(pp, &pn->childq, child) if (NODE_TITLE == pp->node) { pnode_unlink(pp); - break; + return; } - - puts(".Sh SYNOPSIS"); } /* @@ -862,28 +897,24 @@ pnode_printfuncprototype(struct parse *p, struct pnode puts(".Fc"); } -/* TODO: handle "optional" values. */ +/* + * The element is more complicated than it should be because text + * nodes are treated like ".Ar foo", but non-text nodes need to be + * re-sent into the printer (i.e., without the preceding ".Ar"). + * TODO: handle "optional" attribute. + */ static void -pnode_printarg(struct parse *p, struct pnode *pn, int nested) +pnode_printarg(struct parse *p, struct pnode *pn) { struct pnode *pp; - int sv = nested; - if ( ! nested) - fputs(".", stdout); - nested = 1; - TAILQ_FOREACH(pp, &pn->childq, child) - if (NODE_OPTION == pp->node) { - fputs("Fl ", stdout); - pnode_printmacrolinepart(p, pp); - } else if (NODE_TEXT == pp->node) { + TAILQ_FOREACH(pp, &pn->childq, child) { + if (NODE_TEXT == pp->node) { + pnode_printmopen(p); fputs("Ar ", stdout); - pnode_printmacrolinepart(p, pp); - } else if (NODE_ARG == pp->node) - pnode_printarg(p, pp, nested); - - if ( ! sv) - puts(""); + } + pnode_print(p, pp); + } } /* @@ -934,87 +965,95 @@ pnode_print(struct parse *p, struct pnode *pn) { struct pnode *pp; char *cp; - int last; + int last, sv; if (NULL == pn) return; - if (NODE_TEXT != pn->node && NODE_ROOT != pn->node) - printf(".\\\" %s\n", nodes[pn->node].name); + sv = p->newln; switch (pn->node) { case (NODE_ARG): - pnode_printarg(p, pn, 0); + pnode_printarg(p, pn); pnode_unlinksub(pn); break; case (NODE_CITEREFENTRY): + assert(p->newln); pnode_printciterefentry(p, pn); pnode_unlinksub(pn); break; case (NODE_CODE): - fputs(".Li ", stdout); - pnode_printmacroline(p, pn); - pnode_unlinksub(pn); + pnode_printmopen(p); + fputs("Li ", stdout); break; case (NODE_COMMAND): - fputs(".Nm ", stdout); - pnode_printmacroline(p, pn); - pnode_unlinksub(pn); + pnode_printmopen(p); + fputs("Nm ", stdout); break; case (NODE_FUNCTION): - fputs(".Fn ", stdout); - pnode_printmacroline(p, pn); - pnode_unlinksub(pn); + pnode_printmopen(p); + fputs("Fn ", stdout); break; case (NODE_FUNCPROTOTYPE): + assert(p->newln); pnode_printfuncprototype(p, pn); pnode_unlinksub(pn); break; case (NODE_FUNCSYNOPSISINFO): - fputs(".Fd ", stdout); - pnode_printmacroline(p, pn); - pnode_unlinksub(pn); + pnode_printmopen(p); + fputs("Fd ", stdout); break; + case (NODE_OPTION): + pnode_printmopen(p); + fputs("Fl ", stdout); + break; case (NODE_PARA): - /* FIXME: not always. */ + assert(p->newln); puts(".Pp"); break; case (NODE_PARAMETER): - fputs(".Fa \"", stdout); + /* Suppress non-text children... */ + pnode_printmopen(p); + fputs("Fa \"", stdout); pnode_printmacrolinepart(p, pn); puts("\""); pnode_unlinksub(pn); break; case (NODE_PROGRAMLISTING): + assert(p->newln); puts(".Bd -literal"); break; case (NODE_REFMETA): abort(); break; case (NODE_REFNAME): - fputs(".Nm ", stdout); - pnode_printmacroline(p, pn); + /* Suppress non-text children... */ + pnode_printmopen(p); + fputs("Nm ", stdout); + pnode_printmacrolinepart(p, pn); pnode_unlinksub(pn); - return; + break; case (NODE_REFNAMEDIV): + assert(p->newln); puts(".Sh NAME"); break; case (NODE_REFPURPOSE): + assert(p->newln); fputs(".Nd ", stdout); - pnode_printmacroline(p, pn); - pnode_unlinksub(pn); - return; + break; case (NODE_REFSYNOPSISDIV): + assert(p->newln); pnode_printrefsynopsisdiv(p, pn); + puts(".Sh SYNOPSIS"); break; case (NODE_REFSECT1): + assert(p->newln); pnode_printrefsect(p, pn); break; case (NODE_STRUCTNAME): - fputs(".Vt ", stdout); - pnode_printmacroline(p, pn); - pnode_unlinksub(pn); - return; + pnode_printmopen(p); + fputs("Vt ", stdout); + break; case (NODE_TEXT): bufclear(p); bufappend(p, pn); @@ -1024,6 +1063,7 @@ pnode_print(struct parse *p, struct pnode *pn) * XXX: all whitespace, including tabs (?). * Remember to escape control characters and escapes. */ + assert(p->bsz); for (last = '\n', cp = p->b; '\0' != *cp; ) { if ('\n' == last) { /* Consume all whitespace. */ @@ -1039,8 +1079,7 @@ pnode_print(struct parse *p, struct pnode *pn) if ('\\' == last) putchar('e'); } - if ('\n' != last) - putchar('\n'); + p->newln = 0; break; default: break; @@ -1050,8 +1089,22 @@ pnode_print(struct parse *p, struct pnode *pn) pnode_print(p, pp); switch (pn->node) { + case (NODE_ARG): + case (NODE_CODE): + case (NODE_COMMAND): + case (NODE_FUNCTION): + case (NODE_FUNCSYNOPSISINFO): + case (NODE_OPTION): + case (NODE_PARAMETER): + case (NODE_REFNAME): + case (NODE_STRUCTNAME): + case (NODE_TEXT): + pnode_printmclose(p, sv); + break; case (NODE_PROGRAMLISTING): + assert(p->newln); puts(".Ed"); + p->newln = 1; break; default: break; @@ -1089,6 +1142,7 @@ readfile(XML_Parser xp, int fd, * Exit when we've read all or errors have occured * during the parse sequence. */ + p.newln = 1; pnode_printprologue(&p, p.root); pnode_print(&p, p.root); pnode_free(p.root);