Generate C/C++ headers in compile using --include (like xxd) (#596)

Signed-off-by: Juan Cruz Viotti <[email protected]>

Author: jviotti

docs/compile.markdown

@@ -8,7 +8,7 @@ Compiling
 jsonschema compile <schema.json|.yaml> [--http/-h] [--verbose/-v]
   [--resolve/-r <schemas-or-directories> ...] [--extension/-e <extension>]
   [--ignore/-i <schemas-or-directories>] [--fast/-f] [--default-dialect/-d <uri>]
-  [--minify/-m] [--json/-j]
+  [--minify/-m] [--json/-j] [--include/-n <name>]
 ```
 
 The `validate` command will first compile the schema into an optimised
@@ -48,3 +48,9 @@ jsonschema compile path/to/my/schema.json --fast > template.json
 ```sh
 jsonschema compile path/to/my/schema.json --resolve other.json > template.json
 ```
+
+### Compile a JSON Schema to a C/C++ header file
+
+```sh
+jsonschema compile path/to/my/schema.json --include MY_SCHEMA > my_schema.h
+```

src/CMakeLists.txt

@@ -22,6 +22,7 @@ target_link_libraries(jsonschema_cli PRIVATE sourcemeta::core::jsonl)
 target_link_libraries(jsonschema_cli PRIVATE sourcemeta::core::jsonpointer)
 target_link_libraries(jsonschema_cli PRIVATE sourcemeta::core::jsonschema)
 target_link_libraries(jsonschema_cli PRIVATE sourcemeta::core::yaml)
+target_link_libraries(jsonschema_cli PRIVATE sourcemeta::core::regex)
 target_link_libraries(jsonschema_cli PRIVATE sourcemeta::core::alterschema)
 target_link_libraries(jsonschema_cli PRIVATE sourcemeta::core::editorschema)
 target_link_libraries(jsonschema_cli PRIVATE sourcemeta::core::options)

src/command_compile.cc

@@ -1,11 +1,16 @@
 #include <sourcemeta/core/io.h>
 #include <sourcemeta/core/json.h>
 #include <sourcemeta/core/jsonschema.h>
+#include <sourcemeta/core/regex.h>
 #include <sourcemeta/core/yaml.h>
 
 #include <sourcemeta/blaze/compiler.h>
 
-#include <iostream> // std::cerr, std::cout
+#include <algorithm> // std::transform
+#include <cctype>    // std::toupper
+#include <iomanip>   // std::hex, std::setw, std::setfill
+#include <iostream>  // std::cerr, std::cout
+#include <sstream>   // std::ostringstream
 
 #include "command.h"
 #include "configuration.h"
@@ -62,11 +67,67 @@ auto sourcemeta::jsonschema::compile(const sourcemeta::core::Options &options)
   }
 
   const auto template_json{sourcemeta::blaze::to_json(schema_template)};
-  if (options.contains("minify")) {
+
+  if (options.contains("include") && !options.at("include").empty()) {
+    std::string name{options.at("include").front()};
+
+    static const auto IDENTIFIER_PATTERN{
+        sourcemeta::core::to_regex("^[A-Za-z_][A-Za-z0-9_]*$")};
+    if (!IDENTIFIER_PATTERN.has_value() ||
+        !sourcemeta::core::matches(IDENTIFIER_PATTERN.value(), name)) {
+      throw InvalidIncludeIdentifier{name};
+    }
+
+    std::transform(name.begin(), name.end(), name.begin(),
+                   [](unsigned char character) -> unsigned char {
+                     return static_cast<unsigned char>(std::toupper(character));
+                   });
+
+    std::ostringstream json_stream;
+    sourcemeta::core::stringify(template_json, json_stream);
+    const auto json_data{std::move(json_stream).str()};
+
+    constexpr auto BYTES_PER_LINE{16};
+
+    std::cout << "#ifndef SOURCEMETA_JSONSCHEMA_INCLUDE_" << name << "_H_\n";
+    std::cout << "#define SOURCEMETA_JSONSCHEMA_INCLUDE_" << name << "_H_\n";
+    std::cout << "\n";
+    std::cout << "#ifdef __cplusplus\n";
+    std::cout << "#include <string_view>\n";
+    std::cout << "#endif\n";
+    std::cout << "\n";
+    std::cout << "static const char " << name << "_DATA[] =";
+
+    for (std::size_t index = 0; index < json_data.size(); ++index) {
+      if (index % BYTES_PER_LINE == 0) {
+        std::cout << "\n  \"";
+      }
+
+      std::cout << "\\x" << std::hex << std::setw(2) << std::setfill('0')
+                << (static_cast<unsigned int>(
+                       static_cast<unsigned char>(json_data[index])));
+
+      if ((index + 1) % BYTES_PER_LINE == 0 || index + 1 == json_data.size()) {
+        std::cout << "\"";
+      }
+    }
+
+    std::cout << ";\n";
+    std::cout << std::dec;
+    std::cout << "static const unsigned int " << name
+              << "_LENGTH = " << json_data.size() << ";\n";
+    std::cout << "\n";
+    std::cout << "#ifdef __cplusplus\n";
+    std::cout << "static constexpr std::string_view " << name << "{" << name
+              << "_DATA, " << name << "_LENGTH};\n";
+    std::cout << "#endif\n";
+    std::cout << "\n";
+    std::cout << "#endif\n";
+  } else if (options.contains("minify")) {
     sourcemeta::core::stringify(template_json, std::cout);
+    std::cout << "\n";
   } else {
     sourcemeta::core::prettify(template_json, std::cout);
+    std::cout << "\n";
   }
-
-  std::cout << "\n";
 }

src/error.h

@@ -81,6 +81,21 @@ class InvalidLintRuleError : public std::runtime_error {
   std::string rule_;
 };
 
+class InvalidIncludeIdentifier : public std::runtime_error {
+public:
+  InvalidIncludeIdentifier(std::string identifier)
+      : std::runtime_error{"The include identifier is not a valid C/C++ "
+                           "identifier"},
+        identifier_{std::move(identifier)} {}
+
+  [[nodiscard]] auto identifier() const noexcept -> const std::string & {
+    return this->identifier_;
+  }
+
+private:
+  std::string identifier_;
+};
+
 class LintAutoFixError : public std::runtime_error {
 public:
   LintAutoFixError(std::string message, std::filesystem::path path,
@@ -316,6 +331,10 @@ inline auto try_catch(const sourcemeta::core::Options &options,
     const auto is_json{options.contains("json")};
     print_exception(is_json, error);
     return EXIT_FAILURE;
+  } catch (const InvalidIncludeIdentifier &error) {
+    const auto is_json{options.contains("json")};
+    print_exception(is_json, error);
+    return EXIT_FAILURE;
   } catch (const LintAutoFixError &error) {
     const auto is_json{options.contains("json")};
     print_exception(is_json, error);

src/main.cc

@@ -59,8 +59,10 @@ Global Options:
 
    compile <schema.json|.yaml> [--extension/-e <extension>]
            [--ignore/-i <schemas-or-directories>] [--fast/-f] [--minify/-m]
+           [--include/-n <name>]
 
        Compile the given schema into an internal optimised representation.
+       Use --include/-n to output as a C/C++ header file.
 
    test [schemas-or-directories...] [--extension/-e <extension>]
         [--ignore/-i <schemas-or-directories>]
@@ -168,6 +170,7 @@ auto jsonschema_main(const std::string &program, const std::string &command,
   } else if (command == "compile") {
     app.flag("fast", {"f"});
     app.flag("minify", {"m"});
+    app.option("include", {"n"});
     app.parse(argc, argv, {.skip = 1});
     sourcemeta::jsonschema::compile(app);
     return EXIT_SUCCESS;

test/CMakeLists.txt

@@ -359,6 +359,10 @@ add_jsonschema_test_unix(compile/pass_without_extension_json)
 add_jsonschema_test_unix(compile/pass_without_extension_yaml)
 add_jsonschema_test_unix(compile/pass_custom_extension_json)
 add_jsonschema_test_unix(compile/pass_custom_extension_yaml)
+add_jsonschema_test_unix(compile/pass_include)
+add_jsonschema_test_unix(compile/pass_include_lowercase)
+add_jsonschema_test_unix(compile/pass_include_short)
+add_jsonschema_test_unix(compile/fail_include_invalid_identifier)
 
 # Canonicalize
 add_jsonschema_test_unix(canonicalize/pass_input_unmodified)

test/compile/fail_include_invalid_identifier.sh

@@ -0,0 +1,42 @@
+#!/bin/sh
+
+set -o errexit
+set -o nounset
+
+TMP="$(mktemp -d)"
+clean() { rm -rf "$TMP"; }
+trap clean EXIT
+
+cat << 'EOF' > "$TMP/schema.json"
+{
+  "$id": "https://example.com",
+  "$schema": "https://json-schema.org/draft/2020-12/schema",
+  "type": "string"
+}
+EOF
+
+# Test with hyphen (invalid)
+"$1" compile "$TMP/schema.json" --include my-schema 2>"$TMP/stderr.txt" \
+  && CODE="$?" || CODE="$?"
+test "$CODE" = "1" || exit 1
+
+cat << 'EOF' > "$TMP/expected.txt"
+error: The include identifier is not a valid C/C++ identifier
+  at identifier my-schema
+EOF
+
+diff "$TMP/stderr.txt" "$TMP/expected.txt"
+
+# JSON error
+"$1" compile "$TMP/schema.json" --include my-schema --json > "$TMP/stdout.txt" 2>&1 \
+  && CODE="$?" || CODE="$?"
+test "$CODE" = "1" || exit 1
+
+cat << 'EOF' > "$TMP/expected.txt"
+{
+  "error": "The include identifier is not a valid C/C++ identifier",
+  "identifier": "my-schema"
+}
+EOF
+
+diff "$TMP/stdout.txt" "$TMP/expected.txt"

test/compile/pass_include.sh

@@ -0,0 +1,55 @@
+#!/bin/sh
+
+set -o errexit
+set -o nounset
+
+TMP="$(mktemp -d)"
+clean() { rm -rf "$TMP"; }
+trap clean EXIT
+
+cat << 'EOF' > "$TMP/schema.json"
+{
+  "$id": "https://example.com",
+  "$schema": "https://json-schema.org/draft/2020-12/schema",
+  "type": "string"
+}
+EOF
+
+"$1" compile "$TMP/schema.json" --include TEST_SCHEMA > "$TMP/output.h"
+
+cat << 'EOF' > "$TMP/expected.h"
+#ifndef SOURCEMETA_JSONSCHEMA_INCLUDE_TEST_SCHEMA_H_
+#define SOURCEMETA_JSONSCHEMA_INCLUDE_TEST_SCHEMA_H_
+
+#ifdef __cplusplus
+#include <string_view>
+#endif
+
+static const char TEST_SCHEMA_DATA[] =
+  "\x5b\x66\x61\x6c\x73\x65\x2c\x74\x72\x75\x65\x2c\x5b\x22\x22\x2c"
+  "\x22\x68\x74\x74\x70\x73\x3a\x2f\x2f\x65\x78\x61\x6d\x70\x6c\x65"
+  "\x2e\x63\x6f\x6d\x22\x5d\x2c\x5b\x5b\x31\x31\x2c\x22\x2f\x74\x79"
+  "\x70\x65\x22\x2c\x22\x22\x2c\x22\x23\x2f\x74\x79\x70\x65\x22\x2c"
+  "\x32\x2c\x5b\x38\x2c\x34\x5d\x5d\x5d\x5d";
+static const unsigned int TEST_SCHEMA_LENGTH = 74;
+
+#ifdef __cplusplus
+static constexpr std::string_view TEST_SCHEMA{TEST_SCHEMA_DATA, TEST_SCHEMA_LENGTH};
+#endif
+
+#endif
+EOF
+
+diff "$TMP/output.h" "$TMP/expected.h"
+
+# Verify the header compiles with a C compiler
+cat << 'EOF' > "$TMP/test.c"
+#include "output.h"
+EOF
+cc -c "$TMP/test.c" -o "$TMP/test.o"
+
+# Verify the header compiles with a C++ compiler
+cat << 'EOF' > "$TMP/test.cc"
+#include "output.h"
+EOF
+c++ -std=c++17 -c "$TMP/test.cc" -o "$TMP/test_cpp.o"

test/compile/pass_include_lowercase.sh

@@ -0,0 +1,43 @@
+#!/bin/sh
+
+set -o errexit
+set -o nounset
+
+TMP="$(mktemp -d)"
+clean() { rm -rf "$TMP"; }
+trap clean EXIT
+
+cat << 'EOF' > "$TMP/schema.json"
+{
+  "$id": "https://example.com",
+  "$schema": "https://json-schema.org/draft/2020-12/schema",
+  "type": "string"
+}
+EOF
+
+"$1" compile "$TMP/schema.json" --include test_schema > "$TMP/output.h"
+
+cat << 'EOF' > "$TMP/expected.h"
+#ifndef SOURCEMETA_JSONSCHEMA_INCLUDE_TEST_SCHEMA_H_
+#define SOURCEMETA_JSONSCHEMA_INCLUDE_TEST_SCHEMA_H_
+
+#ifdef __cplusplus
+#include <string_view>
+#endif
+
+static const char TEST_SCHEMA_DATA[] =
+  "\x5b\x66\x61\x6c\x73\x65\x2c\x74\x72\x75\x65\x2c\x5b\x22\x22\x2c"
+  "\x22\x68\x74\x74\x70\x73\x3a\x2f\x2f\x65\x78\x61\x6d\x70\x6c\x65"
+  "\x2e\x63\x6f\x6d\x22\x5d\x2c\x5b\x5b\x31\x31\x2c\x22\x2f\x74\x79"
+  "\x70\x65\x22\x2c\x22\x22\x2c\x22\x23\x2f\x74\x79\x70\x65\x22\x2c"
+  "\x32\x2c\x5b\x38\x2c\x34\x5d\x5d\x5d\x5d";
+static const unsigned int TEST_SCHEMA_LENGTH = 74;
+
+#ifdef __cplusplus
+static constexpr std::string_view TEST_SCHEMA{TEST_SCHEMA_DATA, TEST_SCHEMA_LENGTH};
+#endif
+
+#endif
+EOF
+
+diff "$TMP/output.h" "$TMP/expected.h"

test/compile/pass_include_short.sh

@@ -0,0 +1,44 @@
+#!/bin/sh
+
+set -o errexit
+set -o nounset
+
+TMP="$(mktemp -d)"
+clean() { rm -rf "$TMP"; }
+trap clean EXIT
+
+cat << 'EOF' > "$TMP/schema.json"
+{
+  "$id": "https://example.com",
+  "$schema": "https://json-schema.org/draft/2020-12/schema",
+  "type": "string"
+}
+EOF
+
+# Use short option -n
+"$1" compile "$TMP/schema.json" -n TEST_SCHEMA > "$TMP/output.h"
+
+cat << 'EOF' > "$TMP/expected.h"
+#ifndef SOURCEMETA_JSONSCHEMA_INCLUDE_TEST_SCHEMA_H_
+#define SOURCEMETA_JSONSCHEMA_INCLUDE_TEST_SCHEMA_H_
+
+#ifdef __cplusplus
+#include <string_view>
+#endif
+
+static const char TEST_SCHEMA_DATA[] =
+  "\x5b\x66\x61\x6c\x73\x65\x2c\x74\x72\x75\x65\x2c\x5b\x22\x22\x2c"
+  "\x22\x68\x74\x74\x70\x73\x3a\x2f\x2f\x65\x78\x61\x6d\x70\x6c\x65"
+  "\x2e\x63\x6f\x6d\x22\x5d\x2c\x5b\x5b\x31\x31\x2c\x22\x2f\x74\x79"
+  "\x70\x65\x22\x2c\x22\x22\x2c\x22\x23\x2f\x74\x79\x70\x65\x22\x2c"
+  "\x32\x2c\x5b\x38\x2c\x34\x5d\x5d\x5d\x5d";
+static const unsigned int TEST_SCHEMA_LENGTH = 74;
+
+#ifdef __cplusplus
+static constexpr std::string_view TEST_SCHEMA{TEST_SCHEMA_DATA, TEST_SCHEMA_LENGTH};
+#endif
+
+#endif
+EOF
+
+diff "$TMP/output.h" "$TMP/expected.h"