path: root/src/linker.c

#include <assert.h>
#include <errno.h>
#include <limits.h>
#include <string.h>
#include <stdio.h>
#include <stdlib.h>
#include <stdint.h>
#include <sys/stat.h>
#include <libgen.h>

#ifdef WIN32
#include <shlwapi.h>
#endif

#include "jq_parser.h"
#include "locfile.h"
#include "jv.h"
#include "jq.h"
#include "parser.h"
#include "util.h"
#include "compile.h"
#include "jv_alloc.h"

struct lib_loading_state {
  char **names;
  block *defs;
  uint64_t ct;
};
static int load_library(jq_state *jq, jv lib_path,
                        int is_data, int raw, int optional,
                        const char *as, block *out_block,
                        struct lib_loading_state *lib_state);

static int path_is_relative(jv p) {
  const char *s = jv_string_value(p);

#ifdef WIN32
  int res = PathIsRelativeA(s);
#else
  int res = *s != '/';
#endif
  jv_free(p);
  return res;
}


// Given a lib_path to search first, creates a chain of search paths
// in the following order:
// 1. lib_path
// 2. -L paths passed in on the command line (from jq_state*) or builtin list
static jv build_lib_search_chain(jq_state *jq, jv search_path, jv jq_origin, jv lib_origin) {
  assert(jv_get_kind(search_path) == JV_KIND_ARRAY);
  jv expanded = jv_array();
  jv expanded_elt;
  jv err = jv_null();
  jv_array_foreach(search_path, i, path) {
    if (jv_get_kind(path) != JV_KIND_STRING) {
      jv_free(path);
      continue;
    }
    path = expand_path(path);
    if (!jv_is_valid(path)) {
      err = path;
      path = jv_null();
      continue;
    }
    if (strcmp(".",jv_string_value(path)) == 0) {
      expanded_elt = jv_copy(path);
    } else if (strncmp("$ORIGIN/",jv_string_value(path),sizeof("$ORIGIN/") - 1) == 0) {
      expanded_elt = jv_string_fmt("%s/%s",
                               jv_string_value(jq_origin),
                               jv_string_value(path) + sizeof ("$ORIGIN/") - 1);
    } else if (jv_get_kind(lib_origin) == JV_KIND_STRING &&
               path_is_relative(jv_copy(path))) {
      expanded_elt = jv_string_fmt("%s/%s",
                               jv_string_value(lib_origin),
                               jv_string_value(path));
    } else {
      expanded_elt = path;
      path = jv_invalid();
    }
    expanded = jv_array_append(expanded, expanded_elt);
    jv_free(path);
  }
  jv_free(jq_origin);
  jv_free(lib_origin);
  jv_free(search_path);
  return JV_ARRAY(expanded, err);
}

// Doesn't actually check that name not be an absolute path, and we
// don't have to: we always append relative paths to others (with a '/'
// in between).
static jv validate_relpath(jv name) {
  const char *s = jv_string_value(name);
  if (strchr(s, '\\')) {
    jv res = jv_invalid_with_msg(jv_string_fmt("Modules must be named by relative paths using '/', not '\\' (%s)", s));
    jv_free(name);
    return res;
  }
  jv components = jv_string_split(jv_copy(name), jv_string("/"));
  jv_array_foreach(components, i, x) {
    if (!strcmp(jv_string_value(x), "..")) {
      jv_free(x);
      jv_free(components);
      jv res = jv_invalid_with_msg(jv_string_fmt("Relative paths to modules may not traverse to parent directories (%s)", s));
      jv_free(name);
      return res;
    }
    if (i > 0 && jv_equal(jv_copy(x), jv_array_get(jv_copy(components), i - 1))) {
      jv_free(x);
      jv_free(components);
      jv res = jv_invalid_with_msg(jv_string_fmt("module names must not have equal consecutive components: %s",
                                                 jv_string_value(name)));
      jv_free(name);
      return res;
    }
    jv_free(x);
  }
  jv_free(components);
  return name;
}

// Assumes name has been validated
static jv jv_basename(jv name) {
  const char *s = jv_string_value(name);
  const char *p = strrchr(s, '/');
  if (!p)
    return name;
  jv res = jv_string_fmt("%s", p);
  jv_free(name);
  return res;
}

// Asummes validated relative path to module
static jv find_lib(jq_state *jq, jv rel_path, jv search, const char *suffix, jv jq_origin, jv lib_origin) {
  if (!jv_is_valid(rel_path)) {
    jv_free(search);
    jv_free(jq_origin);
    jv_free(lib_origin);
    return rel_path;
  }
  if (jv_get_kind(rel_path) != JV_KIND_STRING) {
    jv_free(rel_path);
    jv_free(search);
    jv_free(jq_origin);
    jv_free(lib_origin);
    return jv_invalid_with_msg(jv_string_fmt("Module path must be a string"));
  }
  if (jv_get_kind(search) != JV_KIND_ARRAY) {
    jv_free(rel_path);
    jv_free(search);
    jv_free(jq_origin);
    jv_free(lib_origin);
    return jv_invalid_with_msg(jv_string_fmt("Module search path must be an array"));
  }

  struct stat st;
  int ret;

  // Ideally we should cache this somewhere
  search = build_lib_search_chain(jq, search, jq_origin, lib_origin);
  jv err = jv_array_get(jv_copy(search), 1);
  search = jv_array_get(search, 0);

  jv bname = jv_basename(jv_copy(rel_path));

  jv_array_foreach(search, i, spath) {
    if (jv_get_kind(spath) == JV_KIND_NULL) {
      jv_free(spath);
      break;
    }
    if (jv_get_kind(spath) != JV_KIND_STRING ||
        strcmp(jv_string_value(spath), "") == 0) {
      jv_free(spath);
      continue; /* XXX report non-strings in search path?? */
    }
    // Try ${search_dir}/${rel_path}.jq
    jv testpath = jq_realpath(jv_string_fmt("%s/%s%s",
                                            jv_string_value(spath),
                                            jv_string_value(rel_path),
                                            suffix));
    ret = stat(jv_string_value(testpath),&st);
    if (ret == -1 && errno == ENOENT) {
      jv_free(testpath);
      // Try ${search_dir}/$(dirname ${rel_path})/jq/main.jq
      testpath = jq_realpath(jv_string_fmt("%s/%s/%s%s",
                                           jv_string_value(spath),
                                           jv_string_value(rel_path),
                                           "jq/main",
                                           suffix));
      ret = stat(jv_string_value(testpath),&st);
    }
    if (ret == -1 && errno == ENOENT) {
      jv_free(testpath);
      // Try ${search_dir}/${rel_path}/$(basename ${rel_path}).jq
      testpath = jq_realpath(jv_string_fmt("%s/%s/%s%s",
                                           jv_string_value(spath),
                                           jv_string_value(rel_path),
                                           jv_string_value(bname),
                                           suffix));
      ret = stat(jv_string_value(testpath),&st);
    }
    if (ret == 0) {
      jv_free(err);
      jv_free(rel_path);
      jv_free(search);
      jv_free(bname);
      jv_free(spath);
      return testpath;
    }
    jv_free(testpath);
    jv_free(spath);
  }
  jv output;
  if (!jv_is_valid(err)) {
    err = jv_invalid_get_msg(err);
    output = jv_invalid_with_msg(jv_string_fmt("module not found: %s (%s)",
                                               jv_string_value(rel_path),