pass-lang/src/lex/indent.c

///
/// See `docs/syntax.md#indentation-levels` for an explanation of the indent level algorithm.
///

#include <stdbool.h>
#include <stdint.h>
#include <stdio.h>
#include <stdlib.h>

#include "../io.h"
#include "indent.h"

_Bool is_indent(char c) {
    return c == ' ' || c == '\t';
}

_Bool is_newline(char c) {
    return c == '\r' || c == '\n';
}

enum indent_type {
    INDENT_TABS,
    INDENT_SPACES,
};

struct indent {
    uint32_t tabs;
    uint32_t spaces;
};

static char good_indent(enum indent_type type) {
    switch (type) {
        case INDENT_TABS:
            return '\t';
        case INDENT_SPACES:
            return ' ';
    }
}

static char bad_indent(enum indent_type type) {
    switch (type) {
        case INDENT_TABS:
            return ' ';
        case INDENT_SPACES:
            return '\t';
    }
}

static uint32_t indent_levels = 0;
static struct indent indents[MAX_INDENTS];
static uint32_t additional_line_length = 0;

static _Bool tabs_allowed(void) {
    return indent_levels == 0 || indents[indent_levels - 1].spaces == 0;
}

// We only throw errors on bad indentation if the line is not empty.
// This function spins to the end of the line to determine whether to throw the error.
static void indent_error(enum indent_type type) {
    char c = peekc();
    while (is_indent(c)) {
        nextc();
        c = peekc();
        if (is_newline(c)) {
            return;
        }
    }
    switch (type) {
        case INDENT_SPACES:
            fprintf(stderr, "lexical error: previous line used spaces at this indentation level; this line used tabs\n");
            exit(1);
        case INDENT_TABS:
            fprintf(stderr, "lexical error: previous lines used tabs at this indentation level; this line used spaces\n");
            exit(1);
    }
}

static void expect_indent(enum indent_type type, uint32_t depth) {
    char good = good_indent(type);
    char bad = bad_indent(type);
    char c = peekc();
    for (uint32_t i = 0; i < depth; i++) {
        if (c == bad) {
            indent_error(type);
            return;
        }
        if (is_newline(c)) {
            return;
        }
        if (c != good) {
            fprintf(stderr, "lexical error: indentation does not match any preceding indentation level\n");
            exit(1);
        }
        nextc();
        c = peekc();
    }
}

static uint32_t count_indents(enum indent_type type) {
    uint32_t counter = 0;
    char indent = good_indent(type);
    char c = peekc();
    while (c == indent) {
        counter++;
        nextc();
        c = peekc();
    }
    return counter;
}

static void new_indent(void) {
    struct indent indent = { 0, 0 };
    indent.tabs = count_indents(INDENT_TABS);
    indent.spaces = count_indents(INDENT_SPACES);
    char c = peekc();
    if (c == '\t' && (indent.spaces > 0 || !tabs_allowed())) {
        fprintf(stderr, "lexical error: all tabs on a line must precede all spaces\n");
        exit(1);
    }
    if (is_newline(c)) {
        return;
    }
    if (indent_levels == MAX_INDENTS) {
        fprintf(stderr, "lexical error: too many indentation levels! factor your code!\n");
        exit(1);
    }
    indents[indent_levels] = indent;
    indent_levels++;
}

int32_t lex_indentation(void) {
    uint32_t indent_level = 0;
    char c = peekc();
    while (true) {
        while (is_newline(c)) {
            nextc();
            c = peekc();
        }
        if (c == 0) {
            indent_levels = 0;
            return indent_level;
        }
        if (!is_indent(c)) {
            break;
        }
        indent_level = 0;
        while (is_indent(c) && indent_level < indent_levels) {
            struct indent indent = indents[indent_level];
            expect_indent(INDENT_TABS, indent.tabs);
            expect_indent(INDENT_SPACES, indent.spaces);
            indent_level++;
            c = peekc();
        }
        if (is_indent(c)) {
            new_indent();
            c = peekc();
            if (!is_newline(c)) {
                indent_levels++;
                return indent_levels;
            }
        }
        c = peekc();
    }
    indent_levels = indent_level;
    return indent_levels;
}