File Coverage

include/eshu_pl.h

Criterion	Covered	Total	%
statement	327	346	94.5
branch	282	358	78.7
condition			n/a
subroutine			n/a
pod			n/a
total	609	704	86.5

/*
 * eshu_pl.h — Perl language indentation scanner
 *
 * Tracks {} () [] nesting depth while skipping strings, heredocs,
 * regex, qw/qq/q constructs, pod sections, and comments.
 * Rewrites leading whitespace only.
 */

#ifndef ESHU_PL_H
#define ESHU_PL_H

#include "eshu.h"

#define ESHU_PL_HEREDOC_MAX  64   /* max length of heredoc terminator */
#define ESHU_PL_QDEPTH_MAX   8   /* max nesting depth for paired delims */

/* ══════════════════════════════════════════════════════════════════
 *  Scanner context — persists across lines
 * ══════════════════════════════════════════════════════════════════ */

typedef struct {
	int             depth;          /* brace/paren/bracket nesting  */
	enum eshu_state state;          /* current scanner state        */
	eshu_config_t   cfg;

	/* Heredoc tracking */
	char            heredoc_tag[ESHU_PL_HEREDOC_MAX];
	int             heredoc_tag_len;
	int             heredoc_indented;  /* <<~ variant */
	int             heredoc_pending;   /* heredoc detected, body starts next line */

	/* Quoted construct tracking (qw, qq, q, s, tr, y, m) */
	char            q_open;         /* opening delimiter */
	char            q_close;        /* closing delimiter (0 for non-paired) */
	int             q_depth;        /* nesting depth for paired delimiters */
	int             q_sections;     /* remaining sections (2 for s///, 1 for others) */

	/* Regex tracking */
	char            rx_delim;       /* regex delimiter char */

	/* Track whether last significant token could precede division */
	int             last_was_value; /* 1 if last token was var/number/)/] */
} eshu_pl_ctx_t;

static void eshu_pl_ctx_init(eshu_pl_ctx_t *ctx, const eshu_config_t *cfg) {
	ctx->depth           = 0;
	ctx->state           = ESHU_CODE;
	ctx->cfg             = *cfg;
	ctx->heredoc_tag[0]  = '\0';
	ctx->heredoc_tag_len = 0;
	ctx->heredoc_indented = 0;
	ctx->heredoc_pending = 0;
	ctx->q_open          = 0;
	ctx->q_close         = 0;
	ctx->q_depth         = 0;
	ctx->q_sections      = 0;
	ctx->rx_delim        = 0;
	ctx->last_was_value  = 0;
}

/* ══════════════════════════════════════════════════════════════════
 *  Delimiter helpers
 * ══════════════════════════════════════════════════════════════════ */

static char eshu_pl_matching_close(char open) {
	switch (open) {
	case '(': return ')';
	case '{': return '}';
	case '[': return ']';
	case '<': return '>';
	default:  return 0; /* non-paired: same char closes */
	}
}

static int eshu_pl_is_paired(char c) {
	return c == '(' || c == '{' || c == '[' || c == '<';
}

/* ══════════════════════════════════════════════════════════════════
 *  Heredoc detection
 *
 *  Recognises:  <
 *  Sets heredoc_pending=1 so the NEXT line enters heredoc state.
 * ══════════════════════════════════════════════════════════════════ */

static int eshu_pl_detect_heredoc(eshu_pl_ctx_t *ctx,
                                  const char *p, const char *end) {
	const char *start;
	int indented = 0;
	char quote = 0;
	int len;

	/* p points to the first '<' — we need "<<" */
	if (p + 1 >= end || *(p + 1) != '<')
		return 0;
	p += 2;

	/* optional ~ for indented heredoc */
	if (p < end && *p == '~') {
		indented = 1;
		p++;
	}

	/* optional quote */
	if (p < end && (*p == '\'' || *p == '"' || *p == '`')) {
		quote = *p;
		p++;
	}

	/* identifier */
	start = p;
	while (p < end && (isalnum((unsigned char)*p) || *p == '_'))
		p++;
	len = (int)(p - start);
	if (len == 0 || len >= ESHU_PL_HEREDOC_MAX)
		return 0;

	/* closing quote must match */
	if (quote && (p >= end || *p != quote))
		return 0;

	memcpy(ctx->heredoc_tag, start, len);
	ctx->heredoc_tag[len]  = '\0';
	ctx->heredoc_tag_len   = len;
	ctx->heredoc_indented  = indented;
	ctx->heredoc_pending   = 1;

	return 1;
}

/* ══════════════════════════════════════════════════════════════════
 *  Check if line is a heredoc terminator
 * ══════════════════════════════════════════════════════════════════ */

static int eshu_pl_is_heredoc_end(const eshu_pl_ctx_t *ctx,
                                  const char *line, const char *eol) {
	const char *p = line;
	int len;

	/* for <<~ the terminator may be indented */
	if (ctx->heredoc_indented) {
		while (p < eol && (*p == ' ' || *p == '\t'))
			p++;
	}

	len = (int)(eol - p);
	/* terminator may have trailing ; or whitespace but we'll be strict:
	   the line content (trimmed) must exactly match the tag */
	if (len < ctx->heredoc_tag_len)
		return 0;

	if (memcmp(p, ctx->heredoc_tag, ctx->heredoc_tag_len) != 0)
		return 0;

	/* rest of line must be empty or just whitespace/semicolons */
	p += ctx->heredoc_tag_len;
	while (p < eol) {
		if (*p != ' ' && *p != '\t' && *p != ';')
			return 0;
		p++;
	}
	return 1;
}

/* ══════════════════════════════════════════════════════════════════
 *  Pod detection — "=word" at start of line
 * ══════════════════════════════════════════════════════════════════ */

static int eshu_pl_is_pod_start(const char *content, const char *eol) {
	if (*content != '=')
		return 0;
	/* must be followed by a letter */
	if (content + 1 >= eol || !isalpha((unsigned char)content[1]))
		return 0;
	/* must NOT be =cut */
	if (eol - content >= 4 && memcmp(content, "=cut", 4) == 0)
		return 0;
	return 1;
}

static int eshu_pl_is_pod_end(const char *content, const char *eol) {
	int len = (int)(eol - content);
	if (len < 4) return 0;
	if (memcmp(content, "=cut", 4) != 0) return 0;
	/* rest should be whitespace or EOL */
	if (len > 4 && content[4] != ' ' && content[4] != '\t')
		return 0;
	return 1;
}

/* ══════════════════════════════════════════════════════════════════
 *  Classify whether a preceding context expects regex or division
 *
 *  Returns 1 if the next '/' should be treated as regex opening.
 * ══════════════════════════════════════════════════════════════════ */

static int eshu_pl_expects_regex(const eshu_pl_ctx_t *ctx) {
	/*
	 * If last token was a "value" (variable, number, closing bracket/paren)
	 * then / is division. Otherwise it's regex.
	 */
	return !ctx->last_was_value;
}

/* ══════════════════════════════════════════════════════════════════
 *  Enter a quoted construct (q/qq/qw/qx/s/tr/y/m)
 *
 *  p points to the character AFTER the keyword letter(s).
 *  Returns the number of chars consumed for the delimiter.
 * ══════════════════════════════════════════════════════════════════ */

static int eshu_pl_enter_quoted(eshu_pl_ctx_t *ctx, char delim,
                                int sections, enum eshu_state state) {
	ctx->q_open     = delim;
	ctx->q_sections = sections;
	if (eshu_pl_is_paired(delim)) {
		ctx->q_close = eshu_pl_matching_close(delim);
		ctx->q_depth = 1;
	} else {
		ctx->q_close = delim;
		ctx->q_depth = 0; /* non-paired don't nest */
	}
	ctx->state = state;
	return 1; /* consumed the delimiter char */
}

/* ══════════════════════════════════════════════════════════════════
 *  Try to detect q/qq/qw/qx/s/tr/y/m constructs
 *
 *  p points to a char that may be q, s, t, m, y.
 *  Returns chars consumed (including keyword + delimiter) or 0.
 * ══════════════════════════════════════════════════════════════════ */

static int eshu_pl_try_q_construct(eshu_pl_ctx_t *ctx,
                                   const char *p, const char *end) {
	const char *start = p;
	char c = *p;
	int sections = 1;
	enum eshu_state st = ESHU_Q;

	if (c == 'q') {
		p++;
		if (p < end && *(p) == 'w') {
			p++; st = ESHU_QW;
		} else if (p < end && *(p) == 'q') {
			p++; st = ESHU_QQ;
		} else if (p < end && *(p) == 'x') {
			p++; st = ESHU_QQ; /* qx behaves like qq */
		} else if (p < end && *(p) == 'r') {
			p++; st = ESHU_QQ; /* qr// behaves like qq for scanning */
		} else {
			st = ESHU_Q;
		}
	} else if (c == 's') {
		p++;
		sections = 2;
		st = ESHU_QQ; /* s/// — two sections, interpolates */
	} else if (c == 't' && p + 1 < end && *(p + 1) == 'r') {
		p += 2;
		sections = 2;
		st = ESHU_QQ;
	} else if (c == 'y') {
		p++;
		sections = 2;
		st = ESHU_QQ;
	} else if (c == 'm') {
		p++;
		st = ESHU_QQ; /* m// uses same delimiter tracking as qq */
	} else {
		return 0;
	}

	/* Must be followed by a non-alnum delimiter */
	if (p >= end)
		return 0;
	if (isalnum((unsigned char)*p) || *p == '_')
		return 0;
	if (*p == ' ' || *p == '\t' || *p == '\n')
		return 0;

	eshu_pl_enter_quoted(ctx, *p, sections, st);
	p++;
	return (int)(p - start);
}

/* Check if the character before position p is a word character
 * (to prevent matching 'eq' as 'q' construct, etc.) */
static int eshu_pl_preceded_by_word(const char *line_start, const char *p) {
	if (p <= line_start)
		return 0;
	return isalnum((unsigned char)*(p - 1)) || *(p - 1) == '_';
}

/* ══════════════════════════════════════════════════════════════════
 *  Scan a line for nesting changes (Perl-aware)
 *
 *  Called AFTER the line has been emitted. Updates ctx->state
 *  and ctx->depth for the next line.
 * ══════════════════════════════════════════════════════════════════ */

static void eshu_pl_scan_line(eshu_pl_ctx_t *ctx,
                              const char *p, const char *end) {
	const char *line_start = p;

	while (p < end) {
		char c = *p;

		switch (ctx->state) {
		case ESHU_CODE:
			if (c == '{' || c == '(' || c == '[') {
				ctx->depth++;
				ctx->last_was_value = 0;
			} else if (c == '}' || c == ')' || c == ']') {
				ctx->depth--;
				if (ctx->depth < 0) ctx->depth = 0;
				ctx->last_was_value = 1;
			} else if (c == '"') {
				ctx->state = ESHU_STRING_DQ;
				ctx->last_was_value = 0;
			} else if (c == '\'') {
				ctx->state = ESHU_STRING_SQ;
				ctx->last_was_value = 0;
			} else if (c == '#') {
				/* line comment — skip rest */
				ctx->last_was_value = 0;
				return;
			} else if (c == '<' && eshu_pl_detect_heredoc(ctx, p, end)) {
				/* heredoc_pending is set; continue scanning rest of line */
				/* skip past the heredoc operator to avoid re-matching */
				p += 2; /* skip << */
				if (p < end && *p == '~') p++;
				if (p < end && (*p == '\'' || *p == '"' || *p == '`')) {
					char hq = *p; p++;
					while (p < end && *p != hq) p++;
					if (p < end) p++;
				} else {
					while (p < end && (isalnum((unsigned char)*p) || *p == '_'))
						p++;
				}
				ctx->last_was_value = 1;
				continue;
			} else if (c == '/' && eshu_pl_expects_regex(ctx)) {
				/* regex literal */
				ctx->rx_delim = '/';
				ctx->state = ESHU_REGEX;
				ctx->last_was_value = 0;
			} else if ((c == 'q' || c == 'm' || c == 's' || c == 'y' ||
			            (c == 't' && p + 1 < end && *(p + 1) == 'r')) &&
			           !eshu_pl_preceded_by_word(line_start, p)) {
				int consumed = eshu_pl_try_q_construct(ctx, p, end);
				if (consumed > 0) {
					p += consumed;
					ctx->last_was_value = 0;
					continue;
				}
				/* not a q-construct, fall through */
				if (isalnum((unsigned char)c))
					ctx->last_was_value = 0;
			} else if (c == '/' && !eshu_pl_expects_regex(ctx)) {
				/* division operator */
				ctx->last_was_value = 0;
			} else if (c == '$' || c == '@' || c == '%') {
				/* variable sigil — skip the variable name */
				ctx->last_was_value = 1;
				p++;
				while (p < end && (isalnum((unsigned char)*p) || *p == '_' || *p == ':'))
					p++;
				continue;
			} else if (isdigit((unsigned char)c)) {
				ctx->last_was_value = 1;
				while (p < end && (isalnum((unsigned char)*p) || *p == '.' || *p == '_'))
					p++;
				continue;
			} else if (c == '=' && p + 1 < end && *(p + 1) == '~') {
				/* =~ forces next / to be regex */
				ctx->last_was_value = 0;
				p += 2;
				continue;
			} else if (c == '!' && p + 1 < end && *(p + 1) == '~') {
				ctx->last_was_value = 0;
				p += 2;
				continue;
			} else if (c == ' ' || c == '\t') {
				/* whitespace — don't change last_was_value */
			} else if (isalpha((unsigned char)c) || c == '_') {
				/* keyword or bareword — skip it */
				const char *ws = p;
				while (p < end && (isalnum((unsigned char)*p) || *p == '_'))
					p++;
				/* barewords ending in a value context: could be function call */
				ctx->last_was_value = 0;
				continue;
			} else {
				/* operator chars: = + - * etc */
				ctx->last_was_value = 0;
			}
			break;

		case ESHU_STRING_DQ:
			if (c == '\\' && p + 1 < end) {
				p++; /* skip escaped char */
			} else if (c == '"') {
				ctx->state = ESHU_CODE;
				ctx->last_was_value = 1;
			}
			break;

		case ESHU_STRING_SQ:
			if (c == '\\' && p + 1 < end) {
				p++; /* skip escaped char */
			} else if (c == '\'') {
				ctx->state = ESHU_CODE;
				ctx->last_was_value = 1;
			}
			break;

		case ESHU_REGEX:
			if (c == '\\' && p + 1 < end) {
				p++; /* skip escaped char */
			} else if (c == ctx->rx_delim) {
				ctx->state = ESHU_CODE;
				ctx->last_was_value = 1;
				/* skip optional flags */
				p++;
				while (p < end && isalpha((unsigned char)*p))
					p++;
				continue;
			}
			break;

		case ESHU_QW:
		case ESHU_QQ:
		case ESHU_Q:
			if (c == '\\' && p + 1 < end && ctx->state != ESHU_QW) {
				p++; /* skip escaped char (not in qw) */
			} else if (ctx->q_close != ctx->q_open) {
				/* paired delimiters — track nesting */
				if (c == ctx->q_open) {
					ctx->q_depth++;
				} else if (c == ctx->q_close) {
					ctx->q_depth--;
					if (ctx->q_depth == 0) {
						ctx->q_sections--;
						if (ctx->q_sections <= 0) {
							ctx->state = ESHU_CODE;
							ctx->last_was_value = 1;
							/* skip optional flags */
							p++;
							while (p < end && isalpha((unsigned char)*p))
								p++;
							continue;
						} else {
							/* next section: find new delimiter */
							p++;
							/* skip whitespace between sections */
							while (p < end && (*p == ' ' || *p == '\t'))
								p++;
							if (p < end) {
								eshu_pl_enter_quoted(ctx, *p,
									ctx->q_sections, ctx->state);
								p++; /* skip past opening delimiter */
							}
							continue;
						}
					}
				}
			} else {
				/* non-paired: same char opens and closes */
				if (c == ctx->q_close) {
					ctx->q_sections--;
					if (ctx->q_sections <= 0) {
						ctx->state = ESHU_CODE;
						ctx->last_was_value = 1;
						/* skip optional flags */
						p++;
						while (p < end && isalpha((unsigned char)*p))
							p++;
						continue;
					}
					/* next section uses same delimiter, just continue */
				}
			}
			break;

		/* These states are handled at the line level, not char level */
		case ESHU_HEREDOC:
		case ESHU_HEREDOC_INDENT:
		case ESHU_POD:
		case ESHU_CHAR_LIT:
		case ESHU_COMMENT_LINE:
		case ESHU_COMMENT_BLOCK:
		case ESHU_PREPROCESSOR:
			return;
		}
		p++;
	}
}

/* ══════════════════════════════════════════════════════════════════
 *  Process a single Perl line — decide indent, emit, scan
 * ══════════════════════════════════════════════════════════════════ */

static void eshu_pl_process_line(eshu_pl_ctx_t *ctx, eshu_buf_t *out,
                                 const char *line_start, const char *eol) {
	const char *content = eshu_skip_leading_ws(line_start);
	int line_len;
	int indent_depth;

	/* empty line — preserve it */
	if (content >= eol) {
		eshu_buf_putc(out, '\n');
		return;
	}

	line_len = (int)(eol - content);

	/* ── Heredoc body: pass through verbatim ── */
	if (ctx->state == ESHU_HEREDOC || ctx->state == ESHU_HEREDOC_INDENT) {
		/* emit line exactly as-is */
		eshu_buf_write_trimmed(out, line_start, (int)(eol - line_start));
		eshu_buf_putc(out, '\n');

		/* check for terminator */
		if (eshu_pl_is_heredoc_end(ctx, line_start, eol)) {
			ctx->state = ESHU_CODE;
			ctx->heredoc_tag[0] = '\0';
			ctx->heredoc_tag_len = 0;
		}
		return;
	}

	/* ── Pod section: pass through verbatim ── */
	if (ctx->state == ESHU_POD) {
		eshu_buf_write_trimmed(out, line_start, (int)(eol - line_start));
		eshu_buf_putc(out, '\n');

		if (eshu_pl_is_pod_end(content, eol))
			ctx->state = ESHU_CODE;
		return;
	}

	/* ── Pod start detection (= at column 0) ── */
	if (content == line_start && eshu_pl_is_pod_start(content, eol)) {
		ctx->state = ESHU_POD;
		eshu_buf_write_trimmed(out, content, line_len);
		eshu_buf_putc(out, '\n');
		return;
	}

	/* ── Inside multi-line q/qq/qw/regex: indent at current depth+1 ── */
	if (ctx->state == ESHU_QW || ctx->state == ESHU_QQ ||
	    ctx->state == ESHU_Q || ctx->state == ESHU_REGEX) {
		int qdepth = ctx->depth + 1;
		/* closing delimiter line gets same depth as opening line */
		if (ctx->q_close && *content == ctx->q_close)
			qdepth = ctx->depth;
		eshu_emit_indent(out, qdepth, &ctx->cfg);
		eshu_buf_write_trimmed(out, content, line_len);
		eshu_buf_putc(out, '\n');
		eshu_pl_scan_line(ctx, content, eol);
		return;
	}

	/* ── Normal Perl code ── */
	indent_depth = ctx->depth;

	/* If line starts with closer, dedent this line */
	if (*content == '}' || *content == ')' || *content == ']') {
		indent_depth--;
		if (indent_depth < 0) indent_depth = 0;
	}

	eshu_emit_indent(out, indent_depth, &ctx->cfg);
	eshu_buf_write_trimmed(out, content, line_len);
	eshu_buf_putc(out, '\n');

	/* scan for nesting changes */
	eshu_pl_scan_line(ctx, content, eol);

	/* If a heredoc was detected on this line, enter heredoc state now */
	if (ctx->heredoc_pending) {
		ctx->heredoc_pending = 0;
		ctx->state = ctx->heredoc_indented ? ESHU_HEREDOC_INDENT : ESHU_HEREDOC;
	}
}

/* ══════════════════════════════════════════════════════════════════
 *  Public API — indent a Perl source string
 * ══════════════════════════════════════════════════════════════════ */

static char * eshu_indent_pl(const char *src, size_t src_len,
                             const eshu_config_t *cfg, size_t *out_len) {
	eshu_pl_ctx_t ctx;
	eshu_buf_t out;
	const char *p   = src;
	const char *end = src + src_len;
	char *result;

	eshu_pl_ctx_init(&ctx, cfg);
	eshu_buf_init(&out, src_len + 256);

	{
		int line_num = 1;
		while (p < end) {
			const char *eol = eshu_find_eol(p);

			if (eshu_in_range(cfg, line_num)) {
				eshu_pl_process_line(&ctx, &out, p, eol);
			} else {
				size_t saved = out.len;
				eshu_pl_process_line(&ctx, &out, p, eol);
				out.len = saved;
				eshu_buf_write_trimmed(&out, p, (int)(eol - p));
				eshu_buf_putc(&out, '\n');
			}

			p = eol;
			if (*p == '\n') p++;
			line_num++;
		}
	}

	/* NUL-terminate */
	eshu_buf_putc(&out, '\0');
	out.len--;

	*out_len = out.len;
	result = out.data;
	return result;
}

#endif /* ESHU_PL_H */