=================================================================== RCS file: /cvs/docbook2mdoc/statistics.c,v retrieving revision 1.3 retrieving revision 1.23 diff -u -p -r1.3 -r1.23 --- docbook2mdoc/statistics.c 2019/04/03 08:39:53 1.3 +++ docbook2mdoc/statistics.c 2019/04/14 12:38:33 1.23 @@ -1,4 +1,4 @@ -/* $Id: statistics.c,v 1.3 2019/04/03 08:39:53 schwarze Exp $ */ +/* $Id: statistics.c,v 1.23 2019/04/14 12:38:33 schwarze Exp $ */ /* * Copyright (c) 2019 Ingo Schwarze * @@ -36,6 +36,16 @@ * Typical usage: * statistics < filenames.txt | sort -n * statistics < filenames.txt | grep '\' | sort -n + * + * Relations already fully implemented are excluded by default. + * The option -a shows all relations. + * + * If two arguments (parent and child) are given, a histogram + * of the number of children of the kind in each parent is given + * in addition to the normal output. + * + * Example usage: + * statistics tgroup colspec < filenames.txt | grep colspec */ struct entry { @@ -52,7 +62,18 @@ static char **stack; static size_t stacksz; static size_t stacki; +static const int nchildsz = 8; +struct nchild { + char *parent; + char *child; + int freq[nchildsz]; + int count; +}; +static struct nchild nchild; +static char *fname; + + /* * Count one instance of a parent-child relation. * Before the special call table_add(NULL, NULL), @@ -70,6 +91,21 @@ table_add(const char *parent, const char *child) return; } + /* Optional parent-child histogram. */ + + if (init_done && parent != NULL && child != NULL && + nchild.parent != NULL && nchild.child != NULL && + strcmp(parent, nchild.parent) == 0 && + strcmp(child, nchild.child) == 0) { + if (nchild.count < nchildsz) { + nchild.freq[nchild.count]++; + if (nchild.count > 0) + nchild.freq[nchild.count - 1]--; + } else if (nchild.count == nchildsz) + puts(fname); + nchild.count++; + } + /* If the table entry already exists, increment its count. */ for (i = 0; i < tablei; i++) { @@ -109,6 +145,9 @@ table_add(const char *parent, const char *child) static void stack_push(const char *name) { + if (nchild.parent != NULL && strcmp(name, nchild.parent) == 0) + nchild.count = 0; + if (stacki == stacksz) { stacksz += 8; stack = reallocarray(stack, stacksz, sizeof(*stack)); @@ -186,13 +225,15 @@ parse_file(int fd, char *fname) continue; } if (in_arg) { - if (in_quotes == 0 && b[pend] == '"') { - in_quotes = 1; + if (in_quotes == 0 && + (b[pend] == '\'' || b[pend] == '"')) { + in_quotes = b[pend] == '"' ? 2 : 1; pend++; continue; } if (advance(b, rlen, &pend, - in_quotes ? "\"" : " >") && rsz > 0) + in_quotes == 2 ? "\"" : + in_quotes == 1 ? "'" : " >") && rsz > 0) break; in_arg = in_quotes = elem_end = 0; if (b[pend] == '>') { @@ -283,7 +324,6 @@ parse_file(int fd, char *fname) int main(int argc, char *argv[]) { - char *fname; size_t fsz, i; ssize_t rsz; int ch, fd, show_all; @@ -298,10 +338,136 @@ main(int argc, char *argv[]) return 1; } } + argc -= optind; + argv += optind; + if (argc > 1) { + nchild.parent = argv[0]; + nchild.child = argv[1]; + } + /* Exclude relations that are already fully implemented. */ if (show_all == 0) { + table_add("ROOT", "refentry"); + table_add("acronym", "TEXT"); + table_add("appendix", NULL); + table_add("article", NULL); + table_add("articleinfo", "pubdate"); + table_add("articleinfo", "title"); + table_add("author", "contrib"); + table_add("author", "email"); + table_add("author", "firstname"); + table_add("author", "othername"); + table_add("author", "surname"); + table_add("authorgroup", "author"); + table_add("authorgroup", "othercredit"); + table_add("blockquote", NULL); + table_add("book", NULL); + table_add("bookinfo", "authorgroup"); + table_add("bookinfo", "legalnotice"); + table_add("bookinfo", "pubdate"); + table_add("bookinfo", "title"); + table_add("chapter", NULL); + table_add("citerefentry", "manvolnum"); + table_add("citerefentry", "refentrytitle"); + table_add("code", "TEXT"); + table_add("computeroutput", "TEXT"); + table_add("constant", "TEXT"); + table_add("date", "TEXT"); + table_add("emphasis", "TEXT"); + table_add("entry", NULL); + table_add("errorname", "TEXT"); + table_add("filename", "TEXT"); + table_add("funcdef", "function"); + table_add("funcdef", "TEXT"); + table_add("funcprototype", "funcdef"); + table_add("funcprototype", "paramdef"); + table_add("funcsynopsis", "funcprototype"); + table_add("funcsynopsis", "funcsynopsisinfo"); + table_add("funcsynopsisinfo", "TEXT"); + table_add("function", "TEXT"); + table_add("glossary", "glossdiv"); + table_add("glossary", "glossentry"); + table_add("glossdef", "para"); + table_add("glossdiv", "glossentry"); + table_add("glossentry", "glossdef"); + table_add("glossentry", "glossterm"); + table_add("glossentry", "indexterm"); + table_add("glosslist", "glossentry"); + table_add("glossterm", "TEXT"); + table_add("indexterm", "primary"); + table_add("indexterm", "secondary"); + table_add("informaltable", "tgroup"); + table_add("itemizedlist", "listitem"); + table_add("legalnotice", NULL); + table_add("link", NULL); + table_add("listitem", NULL); + table_add("literal", "TEXT"); + table_add("literallayout", NULL); + table_add("manvolnum", "TEXT"); + table_add("markup", "TEXT"); + table_add("member", "TEXT"); + table_add("note", NULL); + table_add("orderedlist", "listitem"); + table_add("othercredit", "contrib"); + table_add("othercredit", "email"); + table_add("othercredit", "firstname"); + table_add("othercredit", "othername"); + table_add("othercredit", "surname"); table_add("para", NULL); + table_add("paramdef", "parameter"); + table_add("paramdef", "TEXT"); + table_add("parameter", "TEXT"); + table_add("primary", NULL); + table_add("programlisting", NULL); + table_add("pubdate", "TEXT"); + table_add("refentry", "refentryinfo"); + table_add("refentry", "refmeta"); + table_add("refentry", "refnamediv"); + table_add("refentry", "refsect1"); + table_add("refentry", "refsynopsisdiv"); + table_add("refentrytitle", "TEXT"); + table_add("refmeta", "manvolnum"); + table_add("refmeta", "refentrytitle"); + table_add("refmeta", "refmiscinfo"); + table_add("refmiscinfo", "TEXT"); + table_add("refname", "TEXT"); + table_add("refnamediv", "refname"); + table_add("refnamediv", "refpurpose"); + table_add("refpurpose", "TEXT"); + table_add("refsect1", NULL); + table_add("refsect2", NULL); + table_add("refsynopsisdiv", "funcsynopsis"); + table_add("row", "entry"); + table_add("screen", NULL); + table_add("secondary", NULL); + table_add("section", NULL); + table_add("sect1", NULL); + table_add("sect2", NULL); + table_add("sect3", NULL); + table_add("sect4", NULL); + table_add("sgmltag", "TEXT"); + table_add("simpara", NULL); + table_add("simplelist", "member"); + table_add("structfield", "TEXT"); + table_add("structname", "TEXT"); + table_add("symbol", "TEXT"); + table_add("systemitem", "TEXT"); + table_add("table", "tgroup"); + table_add("table", "title"); + table_add("tbody", "row"); + table_add("term", NULL); + table_add("tgroup", "colspec"); + table_add("tgroup", "tbody"); + table_add("tgroup", "thead"); + table_add("thead", "row"); + table_add("title", "TEXT"); + table_add("type", "TEXT"); + table_add("ulink", NULL); + table_add("userinput", "TEXT"); + table_add("variablelist", "varlistentry"); + table_add("varlistentry", "listitem"); + table_add("varlistentry", "term"); } table_add(NULL, NULL); @@ -329,5 +495,13 @@ main(int argc, char *argv[]) if (table[i].count != -1) printf("%d\t%s\t%s\n", table[i].count, table[i].parent, table[i].child); + + /* Optional parent-child histogram. */ + if (nchild.parent != NULL) { + printf("%s %s", nchild.parent, nchild.child); + for (i = 0; i < nchildsz; i++) + printf(" %d", nchild.freq[i]); + putchar('\n'); + } return 0; }