=================================================================== RCS file: /cvs/docbook2mdoc/statistics.c,v retrieving revision 1.1 retrieving revision 1.21 diff -u -p -r1.1 -r1.21 --- docbook2mdoc/statistics.c 2019/03/29 15:55:28 1.1 +++ docbook2mdoc/statistics.c 2019/04/13 13:06:35 1.21 @@ -1,4 +1,4 @@ -/* $Id: statistics.c,v 1.1 2019/03/29 15:55:28 schwarze Exp $ */ +/* $Id: statistics.c,v 1.21 2019/04/13 13:06:35 schwarze Exp $ */ /* * Copyright (c) 2019 Ingo Schwarze * @@ -18,6 +18,7 @@ #include #include #include +#include #include #include #include @@ -35,6 +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 { @@ -51,21 +62,59 @@ 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), + * mark relations to not be counted; + * in that phase, child can be NULL as a wildcard. */ static void table_add(const char *parent, const char *child) { - size_t i; + static int init_done; + size_t i; + if (parent == NULL && child == NULL) { + init_done = 1; + 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++) { if (strcmp(parent, table[i].parent) == 0 && - strcmp(child, table[i].child) == 0) { - table[i].count++; + (child == NULL || table[i].child == NULL || + strcmp(child, table[i].child) == 0)) { + assert(init_done); + if (table[i].count != -1) + table[i].count++; return; } } @@ -83,9 +132,11 @@ table_add(const char *parent, const char *child) if ((table[tablei].parent = strdup(parent)) == NULL) err(1, NULL); - if ((table[tablei].child = strdup(child)) == NULL) + if (child == NULL) + table[tablei].child = NULL; + else if ((table[tablei].child = strdup(child)) == NULL) err(1, NULL); - table[tablei++].count = 1; + table[tablei++].count = init_done ? 1 : -1; } /* @@ -94,6 +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)); @@ -150,6 +204,7 @@ static void parse_file(int fd, char *fname) { char b[4096]; + char *cp; ssize_t rsz; /* Return value from read(2). */ size_t rlen; /* Number of bytes in b[]. */ size_t poff; /* Parse offset in b[]. */ @@ -170,13 +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] == '>') { @@ -217,6 +274,18 @@ parse_file(int fd, char *fname) } else if (b[poff] == '<') { if (advance(b, rlen, &pend, " >") && rsz > 0) break; + if (pend > poff + 3 && + strncmp(b + poff, ""); + if (cp == NULL) { + pend = rlen; + if (rsz > 0) + break; + } else + pend = cp + 3 - b; + continue; + } elem_end = 0; if (b[pend] != '>') in_tag = 1; @@ -232,7 +301,7 @@ parse_file(int fd, char *fname) poff++; } else if (b[poff] != '!' && b[poff] != '?') { table_add(stacki > 0 ? - stack[stacki - 1] : "", + stack[stacki - 1] : "ROOT", b + poff); stack_push(b + poff); } @@ -255,15 +324,144 @@ parse_file(int fd, char *fname) int main(int argc, char *argv[]) { - char *fname; size_t fsz, i; ssize_t rsz; - int fd; + int ch, fd, show_all; - fd = -1; - fname = NULL; + show_all = 0; + while ((ch = getopt(argc, argv, "a")) != -1) { + switch (ch) { + case 'a': + show_all = 1; + break; + default: + 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("blockquote", NULL); + table_add("book", NULL); + table_add("bookinfo", "pubdate"); + table_add("bookinfo", "title"); + table_add("chapter", NULL); + 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("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", "refmeta"); + table_add("refentry", "refnamediv"); + table_add("refentry", "refsect1"); + table_add("refentry", "refsynopsisdiv"); + table_add("refmeta", "manvolnum"); + table_add("refmeta", "refentrytitle"); + 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("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); + /* Loop over input files. */ + fd = -1; + fname = NULL; while ((rsz = getline(&fname, &fsz, stdin)) != -1) { if (fname[rsz - 1] == '\n') fname[--rsz] = '\0'; @@ -282,7 +480,16 @@ main(int argc, char *argv[]) /* Dump results. */ for (i = 0; i < tablei; i++) - printf("%d\t%s\t%s\n", table[i].count, - table[i].parent, table[i].child); + 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; }