perl.c

/*
*   $Id$
*
*   Copyright (c) 2000-2003, Darren Hiebert
*
*   This source code is released for free distribution under the terms of the
*   GNU General Public License.
*
*   This module contains functions for generating tags for PERL language
*   files.
*/

/*
*   INCLUDE FILES
*/
#include "general.h"  /* must always come first */
#include "debug.h"

#include <string.h>

#include "entry.h"
#include "options.h"
#include "read.h"
#include "routines.h"
#include "vstring.h"

#define TRACE_PERL_C 0
#define TRACE if (TRACE_PERL_C) printf("perl.c:%d: ", __LINE__), printf

/*
*   DATA DEFINITIONS
*/
typedef enum {
	K_NONE = -1,
	K_CONSTANT,
	K_FORMAT,
	K_LABEL,
	K_PACKAGE,
	K_SUBROUTINE,
	K_SUBROUTINE_DECLARATION
} perlKind;

static kindOption PerlKinds [] = {
	{ TRUE,  'c', "constant",               "constants" },
	{ TRUE,  'f', "format",                 "formats" },
	{ TRUE,  'l', "label",                  "labels" },
	{ TRUE,  'p', "package",                "packages" },
	{ TRUE,  's', "subroutine",             "subroutines" },
	{ FALSE, 'd', "subroutine declaration", "subroutine declarations" },
};

/*
*   FUNCTION DEFINITIONS
*/

static boolean isIdentifier1 (int c)
{
	return (boolean) (isalpha (c) || c == '_');
}

static boolean isIdentifier (int c)
{
	return (boolean) (isalnum (c) || c == '_');
}

static boolean isPodWord (const char *word)
{
	/* Perl POD words are three to five characters in size.  We use this
	 * fact to find (or not find) the right side of the word and then
	 * perform comparisons, if necessary, of POD words of that size.
	 */
	size_t len;
	for (len = 0; len < 6; ++len)
		if ('\0' == word[len] || ' ' == word[len] || '\t' == word[len])
			break;
	switch (len) {
		case 3:
			return 0 == strncmp(word, "end", 3)
				|| 0 == strncmp(word, "for", 3)
				|| 0 == strncmp(word, "pod", 3);
		case 4:
			return 0 == strncmp(word, "back", 4)
				|| 0 == strncmp(word, "item", 4)
				|| 0 == strncmp(word, "over", 4);
		case 5:
			return 0 == strncmp(word, "begin", 5)
				|| 0 == strncmp(word, "head1", 5)
				|| 0 == strncmp(word, "head2", 5)
				|| 0 == strncmp(word, "head3", 5)
				|| 0 == strncmp(word, "head4", 5);
		default:
			return FALSE;
	}
}

/*
 * Perl subroutine declaration may look like one of the following:
 *
 *  sub abc;
 *  sub abc :attr;
 *  sub abc (proto);
 *  sub abc (proto) :attr;
 *
 * Note that there may be more than one attribute.  Attributes may
 * have things in parentheses (they look like arguments).  Anything
 * inside of those parentheses goes.  Prototypes may contain semi-colons.
 * The matching end when we encounter (outside of any parentheses) either
 * a semi-colon (that'd be a declaration) or an left curly brace
 * (definition).
 *
 * This is pretty complicated parsing (plus we all know that only perl can
 * parse Perl), so we are only promising best effort here.
 *
 * If we can't determine what this is (due to a file ending, for example),
 * we will return FALSE.
 */
static boolean isSubroutineDeclaration (const unsigned char *cp)
{
	boolean attr = FALSE;
	int nparens = 0;

	do {
		for ( ; *cp; ++cp) {
SUB_DECL_SWITCH:
			switch (*cp) {
				case ':':
					if (nparens)
						break;
					else if (TRUE == attr)
						return FALSE;    /* Invalid attribute name */
					else
						attr = TRUE;
					break;
				case '(':
					++nparens;
					break;
				case ')':
					--nparens;
					break;
				case ' ':
				case '\t':
					break;
				case ';':
					if (!nparens)
						return TRUE;
				case '{':
					if (!nparens)
						return FALSE;
				default:
					if (attr) {
						if (isIdentifier1(*cp)) {
							cp++;
							while (isIdentifier (*cp))
								cp++;
							attr = FALSE;
							goto SUB_DECL_SWITCH; /* Instead of --cp; */
						} else {
							return FALSE;
						}
					} else if (nparens) {
						break;
					} else {
						return FALSE;
					}
			}
		}
	} while (NULL != (cp = fileReadLine ()));

	return FALSE;
}

/* `end' points to the equal sign.  Parse from right to left to get the
 * identifier.  Assume we're dealing with something of form \s*\w+\s*=>
 */
static void makeTagFromLeftSide (const char *begin, const char *end,
	vString *name, vString *package)
{
	tagEntryInfo entry;
	const char *b, *e;
	for (e = end - 1; e > begin && isspace(*e); --e)
		;
	if (e < begin)
		return;
	for (b = e; b >= begin && isIdentifier(*b); --b)
		;
	/* Identifier must be either beginning of line of have some whitespace
	 * on its left:
	 */
	if (b < begin || isspace(*b) || ',' == *b)
		++b;
	else if (b != begin)
		return;
	Assert(e - b + 1 > 0);
	vStringClear(name);
	vStringNCatS(name, b, e - b + 1);
	initTagEntry(&entry, vStringValue(name));
	entry.kind = PerlKinds[K_CONSTANT].letter;
	entry.kindName = PerlKinds[K_CONSTANT].name;
	makeTagEntry(&entry);
	if (Option.include.qualifiedTags && package && vStringLength(package)) {
		vStringClear(name);
		vStringCopy(name, package);
		vStringNCatS(name, b, e - b + 1);
		initTagEntry(&entry, vStringValue(name));
		entry.kind = PerlKinds[K_CONSTANT].letter;
		entry.kindName = PerlKinds[K_CONSTANT].name;
		makeTagEntry(&entry);
	}
}

enum const_state { CONST_STATE_NEXT_LINE, CONST_STATE_HIT_END };

/* Parse a single line, find as many NAME => VALUE pairs as we can and try
 * to detect the end of the hashref.
 */
static enum const_state parseConstantsFromLine (const char *cp,
	vString *name, vString *package)
{
	while (1) {
		const size_t sz = strcspn(cp, "#}=");
		switch (cp[sz]) {
			case '=':
				if (cp[sz + 1] && '>' == cp[sz + 1])
					makeTagFromLeftSide(cp, cp + sz, name, package);
				break;
			case '}':	/* Assume this is the end of the hashref. */
				return CONST_STATE_HIT_END;
			case '\0':	/* End of the line. */
			case '#':	/* Assume this is a comment and thus end of the line. */
				return CONST_STATE_NEXT_LINE;
		}
		cp += sz + 1;
	}
}

/* Parse constants declared via hash reference, like this:
 * use constant {
 *   A => 1,
 *   B => 2,
 * };
 * The approach we take is simplistic, but it covers the vast majority of
 * cases well.  There can be some false positives.
 * Returns 0 if found the end of the hashref, -1 if we hit EOF
 */
static int parseConstantsFromHashRef (const unsigned char *cp,
	vString *name, vString *package)
{
	while (1) {
		enum const_state state =
			parseConstantsFromLine((const char *) cp, name, package);
		switch (state) {
			case CONST_STATE_NEXT_LINE:
				cp = fileReadLine();
				if (cp)
					break;
				else
					return -1;
			case CONST_STATE_HIT_END:
				return 0;
		}
	}
}

/* Algorithm adapted from from GNU etags.
 * Perl support by Bart Robinson <lomew@cs.utah.edu>
 * Perl sub names: look for /^ [ \t\n]sub [ \t\n]+ [^ \t\n{ (]+/
 */
static void findPerlTags (void)
{
	vString *name = vStringNew ();
	vString *package = NULL;
	boolean skipPodDoc = FALSE;
	const unsigned char *line;

	while ((line = fileReadLine ()) != NULL)
	{
		boolean spaceRequired = FALSE;
		boolean qualified = FALSE;
		const unsigned char *cp = line;
		perlKind kind = K_NONE;
		tagEntryInfo e;

		if (skipPodDoc)
		{
			if (strncmp ((const char*) line, "=cut", (size_t) 4) == 0)
				skipPodDoc = FALSE;
			continue;
		}
		else if (line [0] == '=')
		{
			skipPodDoc = isPodWord ((const char*)line + 1);
			continue;
		}
		else if (strcmp ((const char*) line, "__DATA__") == 0)
			break;
		else if (strcmp ((const char*) line, "__END__") == 0)
			break;
		else if (line [0] == '#')
			continue;

		while (isspace (*cp))
			cp++;

		if (strncmp((const char*) cp, "sub", (size_t) 3) == 0)
		{
			TRACE("this looks like a sub\n");
			cp += 3;
			kind = K_SUBROUTINE;
			spaceRequired = TRUE;
			qualified = TRUE;
		}
		else if (strncmp((const char*) cp, "use", (size_t) 3) == 0)
		{
			cp += 3;
			if (!isspace(*cp))
				continue;
			while (*cp && isspace (*cp))
				++cp;
			if (strncmp((const char*) cp, "constant", (size_t) 8) != 0)
				continue;
			cp += 8;
			/* Skip up to the first non-space character, skipping empty
			 * and comment lines.
			 */
			while (isspace(*cp))
				cp++;
			while (!*cp || '#' == *cp) {
				cp = fileReadLine ();
				if (!cp)
					goto END_MAIN_WHILE;
				while (isspace (*cp))
					cp++;
			}
			if ('{' == *cp) {
				++cp;
				if (0 == parseConstantsFromHashRef(cp, name, package)) {
					vStringClear(name);
					continue;
				} else
					goto END_MAIN_WHILE;
			}
			kind = K_CONSTANT;
			spaceRequired = FALSE;
			qualified = TRUE;
		}
		else if (strncmp((const char*) cp, "package", (size_t) 7) == 0)
		{
			/* This will point to space after 'package' so that a tag
			   can be made */
			const unsigned char *space = cp += 7;

			if (package == NULL)
				package = vStringNew ();
			else
				vStringClear (package);
			while (isspace (*cp))
				cp++;
			while (*cp && (int) *cp != ';'  &&  !isspace ((int) *cp))
			{
				vStringPut (package, (int) *cp);
				cp++;
			}
			vStringCatS (package, "::");

			cp = space;	 /* Rewind */
			kind = K_PACKAGE;
			spaceRequired = TRUE;
			qualified = TRUE;
		}
		else if (strncmp((const char*) cp, "format", (size_t) 6) == 0)
		{
			cp += 6;
			kind = K_FORMAT;
			spaceRequired = TRUE;
			qualified = TRUE;
		}
		else
		{
			if (isIdentifier1 (*cp))
			{
				const unsigned char *p = cp;
				while (isIdentifier (*p))
					++p;
				while (isspace (*p))
					++p;
				if ((int) *p == ':' && (int) *(p + 1) != ':')
					kind = K_LABEL;
			}
		}
		if (kind != K_NONE)
		{
			TRACE("cp0: %s\n", (const char *) cp);
			if (spaceRequired && *cp && !isspace (*cp))
				continue;

			TRACE("cp1: %s\n", (const char *) cp);
			while (isspace (*cp))
				cp++;

			while (!*cp || '#' == *cp) { /* Gobble up empty lines
				                            and comments */
				cp = fileReadLine ();
				if (!cp)
					goto END_MAIN_WHILE;
				while (isspace (*cp))
					cp++;
			}

			while (isIdentifier (*cp) || (K_PACKAGE == kind && ':' == *cp))
			{
				vStringPut (name, (int) *cp);
				cp++;
			}

			if (K_FORMAT == kind &&
				vStringLength (name) == 0 && /* cp did not advance */
				'=' == *cp)
			{
				/* format's name is optional.  If it's omitted, 'STDOUT'
				   is assumed. */
				vStringCatS (name, "STDOUT");
			}

			vStringTerminate (name);
			TRACE("name: %s\n", name->buffer);

			if (0 == vStringLength(name)) {
				vStringClear(name);
				continue;
			}

			if (K_SUBROUTINE == kind)
			{
				/*
				 * isSubroutineDeclaration() may consume several lines.  So
				 * we record line positions.
				 */
				initTagEntry(&e, vStringValue(name));

				if (TRUE == isSubroutineDeclaration(cp)) {
					if (TRUE == PerlKinds[K_SUBROUTINE_DECLARATION].enabled) {
						kind = K_SUBROUTINE_DECLARATION;
					} else {
						vStringClear (name);
						continue;
					}
				}

				e.kind     = PerlKinds[kind].letter;
				e.kindName = PerlKinds[kind].name;

				makeTagEntry(&e);

				if (Option.include.qualifiedTags && qualified &&
					package != NULL  && vStringLength (package) > 0)
				{
					vString *const qualifiedName = vStringNew ();
					vStringCopy (qualifiedName, package);
					vStringCat (qualifiedName, name);
					e.name = vStringValue(qualifiedName);
					makeTagEntry(&e);
					vStringDelete (qualifiedName);
				}
			} else if (vStringLength (name) > 0)
			{
				makeSimpleTag (name, PerlKinds, kind);
				if (Option.include.qualifiedTags && qualified &&
					K_PACKAGE != kind &&
					package != NULL  && vStringLength (package) > 0)
				{
					vString *const qualifiedName = vStringNew ();
					vStringCopy (qualifiedName, package);
					vStringCat (qualifiedName, name);
					makeSimpleTag (qualifiedName, PerlKinds, kind);
					vStringDelete (qualifiedName);
				}
			}
			vStringClear (name);
		}
	}

END_MAIN_WHILE:
	vStringDelete (name);
	if (package != NULL)
		vStringDelete (package);
}

extern parserDefinition* PerlParser (void)
{
	static const char *const extensions [] = { "pl", "pm", "plx", "perl", NULL };
	parserDefinition* def = parserNew ("Perl");
	def->kinds      = PerlKinds;
	def->kindCount  = KIND_COUNT (PerlKinds);
	def->extensions = extensions;
	def->parser     = findPerlTags;
	return def;
}

/* vi:set tabstop=4 shiftwidth=4 noexpandtab: */