From 7d90f0b520fda6cd9def6cd248c54b075717e948 Mon Sep 17 00:00:00 2001 From: Carl Woffenden Date: Wed, 19 Jan 2022 11:32:53 +0100 Subject: [PATCH] Test and tidy Made the Python more Python-like. Added notes and general tidy. Tested exclusions and building with various options. Tested all scripts. --- build/single_file_libs/build_library_test.sh | 2 +- build/single_file_libs/combine.py | 232 ++++++++++-------- build/single_file_libs/combine.sh | 1 + .../create_single_file_decoder.sh | 6 +- .../create_single_file_library.sh | 6 +- build/single_file_libs/zstd-in.c | 2 +- build/single_file_libs/zstddeclib-in.c | 2 +- 7 files changed, 133 insertions(+), 118 deletions(-) diff --git a/build/single_file_libs/build_library_test.sh b/build/single_file_libs/build_library_test.sh index 7fb99656..31545fc3 100755 --- a/build/single_file_libs/build_library_test.sh +++ b/build/single_file_libs/build_library_test.sh @@ -69,7 +69,7 @@ fi echo "Single file library creation script: PASSED" # Copy the header to here (for the tests) -cp "$ZSTD_SRC_ROOT/zstd.h" zstd.h +cp "$ZSTD_SRC_ROOT/zstd.h" examples/zstd.h # Compile the generated output cc -Wall -Wextra -Werror -Wshadow -pthread -I. -Os -g0 -o $OUT_FILE zstd.c examples/roundtrip.c diff --git a/build/single_file_libs/combine.py b/build/single_file_libs/combine.py index 00a6f11f..0538ccb6 100755 --- a/build/single_file_libs/combine.py +++ b/build/single_file_libs/combine.py @@ -2,6 +2,18 @@ # Tool to bundle multiple C/C++ source files, inlining any includes. # +# Note: there are two types of exclusion options: the '-x' flag, which besides +# excluding a file also adds an #error directive in place of the #include, and +# the '-k' flag, which keeps the #include and doesn't inline the file. The +# intended use cases are: '-x' for files that would normally be #if'd out, so +# features that 100% won't be used in the amalgamated file, for which every +# occurrence adds the error, and '-k' for headers that we wish to manually +# include, such as a project's public API, for which occurrences after the first +# are removed. +# +# Todo: the error handling could be better, which currently throws and halts +# (which is functional just not very friendly). +# # Author: Carl Woffenden, Numfum GmbH (this script is released under a CC0 license/Public Domain) import argparse, re, sys @@ -31,18 +43,18 @@ found: Set[Path] = set() # Compiled regex Patern to handle the following type of file includes: # -# #include "file" -# #include "file" -# # include "file" -# #include "file" -# #include "file" // comment -# #include "file" // comment with quote " +# #include "file" +# #include "file" +# # include "file" +# #include "file" +# #include "file" // comment +# #include "file" // comment with quote " # # And all combinations of, as well as ignoring the following: # -# #include -# //#include "file" -# /*#include "file"*/ +# #include +# //#include "file" +# /*#include "file"*/ # # We don't try to catch errors since the compiler will do this (and the code is # expected to be valid before processing) and we don't care what follows the @@ -54,27 +66,27 @@ include_regex: Pattern = re.compile(r'^\s*#\s*include\s*"(.+?)"') # Simple tests to prove include_regex's cases. # def test_match_include() -> bool: - if (include_regex.match('#include "file"') and - include_regex.match(' #include "file"') and - include_regex.match('# include "file"') and - include_regex.match('#include "file"') and - include_regex.match('#include "file" // comment')): - if (not include_regex.match('#include ') and - not include_regex.match('//#include "file"') and - not include_regex.match('/*#include "file"*/')): - found = include_regex.match('#include "file" // "') - if (found and found.group(1) == 'file'): - print('#include match valid') - return True - return False + if (include_regex.match('#include "file"') and + include_regex.match(' #include "file"') and + include_regex.match('# include "file"') and + include_regex.match('#include "file"') and + include_regex.match('#include "file" // comment')): + if (not include_regex.match('#include ') and + not include_regex.match('//#include "file"') and + not include_regex.match('/*#include "file"*/')): + found = include_regex.match('#include "file" // "') + if (found and found.group(1) == 'file'): + print('#include match valid') + return True + return False # Compiled regex Patern to handle "#pragma once" in various formats: # -# #pragma once -# #pragma once -# # pragma once -# #pragma once -# #pragma once // comment +# #pragma once +# #pragma once +# # pragma once +# #pragma once +# #pragma once // comment # # Ignoring commented versions, same as include_regex. # @@ -83,103 +95,105 @@ pragma_regex: Pattern = re.compile(r'^\s*#\s*pragma\s*once\s*') # Simple tests to prove pragma_regex's cases. # def text_match_pragma() -> bool: - if (pragma_regex.match('#pragma once') and - pragma_regex.match(' #pragma once') and - pragma_regex.match('# pragma once') and - pragma_regex.match('#pragma once') and - pragma_regex.match('#pragma once // comment')): - if (not pragma_regex.match('//#pragma once') and - not pragma_regex.match('/*#pragma once*/')): - print('#pragma once match valid') - return True - return False + if (pragma_regex.match('#pragma once') and + pragma_regex.match(' #pragma once') and + pragma_regex.match('# pragma once') and + pragma_regex.match('#pragma once') and + pragma_regex.match('#pragma once // comment')): + if (not pragma_regex.match('//#pragma once') and + not pragma_regex.match('/*#pragma once*/')): + print('#pragma once match valid') + return True + return False # Finds 'file'. First the currently processing file's 'parent' path is looked at # for a match, followed by the list of 'root' paths, returning a valid Path in # canonical form. If no match is found None is returned. # def resolve_include(file: str, parent: Optional[Path] = None) -> Optional[Path]: - if (parent): - found = parent.joinpath(file).resolve(); - else: - found = Path(file) - if (found.is_file()): - return found - for root in roots: - found = root.joinpath(file).resolve() - if (found.is_file()): - return found - return None + if (parent): + found = parent.joinpath(file).resolve(); + else: + found = Path(file) + if (found.is_file()): + return found + for root in roots: + found = root.joinpath(file).resolve() + if (found.is_file()): + return found + return None # Helper to resolve lists of files. 'file_list' is passed in from the arguments # and each entry resolved to its canonical path (like any include entry, either # from the list of root paths or the owning file's 'parent', which in this case # is case is the input file). The results are stored in 'resolved'. # -def resolve_files(file_list: Optional[List[str]], resolved: Set[Path], parent: Optional[Path] = None) -> None: - if (file_list): - for filename in file_list: - found = resolve_include(filename, parent) - if (found): - resolved.add(found) - else: - error_line(f'Warning: excluded file not found: {filename}') +def resolve_excluded_files(file_list: Optional[List[str]], resolved: Set[Path], parent: Optional[Path] = None) -> None: + if (file_list): + for filename in file_list: + found = resolve_include(filename, parent) + if (found): + resolved.add(found) + else: + error_line(f'Warning: excluded file not found: {filename}') # Writes 'line' to the open 'destn' (or stdout). # def write_line(line: str) -> None: - print(line, file=destn) + print(line, file=destn) # Logs 'line' to stderr. This is also used for general notifications that we # don't want to go to stdout (so the source can be piped). # def error_line(line: Any) -> None: - print(line, file=sys.stderr) + print(line, file=sys.stderr) # Inline the contents of 'file' (with any of its includes also inlined, etc.). # -def add_file(file: Path) -> None: - if (file.is_file()): - error_line(f'Processing: {file}') - with file.open('r') as opened: - for line in opened: - line = line.rstrip('\n') - match_include = include_regex.match(line); - if (match_include): - # We have a quoted include directive so grab the file - inc_name = match_include.group(1) - resolved = resolve_include(inc_name, file.parent) - if (resolved): - if (resolved in excludes): - # The file was excluded so error if the source attempts to use it - write_line(f'#error Using excluded file: {inc_name}') - error_line(f'Excluding: {inc_name}') - else: - if (resolved not in found): - # The file was not previously encountered - found.add(resolved) - if (resolved in keeps): - # But the include was flagged to keep as included - write_line(f'/**** *NOT* inlining {inc_name} ****/') - write_line(line) - error_line('Not Inlining: {inc_name}') - else: - # The file was neither excluded nor seen before so inline it - write_line(f'/**** start inlining {inc_name} ****/') - add_file(resolved) - write_line(f'/**** ended inlining {inc_name} ****/') - else: - write_line(f'/**** skipping file: {inc_name} ****/') - else: - # The include file didn't resolve to a file - write_line(f'#error Unable to find: {inc_name}') - error_line(f'Error: Unable to find: {inc_name}') - else: - # Skip any 'pragma once' directives, otherwise write the source line - if (keep_pragma or not pragma_regex.match(line)): - write_line(line) - else: - error_line(f'Error: Invalid file: {file}') +def add_file(file: Path, file_name: str = None) -> None: + if (file.is_file()): + if (not file_name): + file_name = file.name + error_line(f'Processing: {file_name}') + with file.open('r') as opened: + for line in opened: + line = line.rstrip('\n') + match_include = include_regex.match(line); + if (match_include): + # We have a quoted include directive so grab the file + inc_name = match_include.group(1) + resolved = resolve_include(inc_name, file.parent) + if (resolved): + if (resolved in excludes): + # The file was excluded so error if the compiler uses it + write_line(f'#error Using excluded file: {inc_name}') + error_line(f'Excluding: {inc_name}') + else: + if (resolved not in found): + # The file was not previously encountered + found.add(resolved) + if (resolved in keeps): + # But the include was flagged to keep as included + write_line(f'/**** *NOT* inlining {inc_name} ****/') + write_line(line) + error_line(f'Not inlining: {inc_name}') + else: + # The file was neither excluded nor seen before so inline it + write_line(f'/**** start inlining {inc_name} ****/') + add_file(resolved, inc_name) + write_line(f'/**** ended inlining {inc_name} ****/') + else: + write_line(f'/**** skipping file: {inc_name} ****/') + else: + # The include file didn't resolve to a file + write_line(f'#error Unable to find: {inc_name}') + error_line(f'Error: Unable to find: {inc_name}') + else: + # Skip any 'pragma once' directives, otherwise write the source line + if (keep_pragma or not pragma_regex.match(line)): + write_line(line) + else: + error_line(f'Error: Invalid file: {file}') # Start here parser = argparse.ArgumentParser(description='Amalgamate Tool', epilog=f'example: {sys.argv[0]} -r ../my/path -r ../other/path -o out.c in.c') @@ -197,19 +211,19 @@ found.add(args.input) # Resolve all of the root paths upfront (we'll halt here on invalid roots) if (args.root): - for path in args.root: - roots.add(path.resolve(strict=True)) + for path in args.root: + roots.add(path.resolve(strict=True)) # The remaining params: so resolve the excluded files and #pragma once directive -resolve_files(args.exclude, excludes, args.input.parent) -resolve_files(args.keep, keeps, args.input.parent) +resolve_excluded_files(args.exclude, excludes, args.input.parent) +resolve_excluded_files(args.keep, keeps, args.input.parent) keep_pragma = args.pragma; # Then recursively process the input file try: - if (args.output): - destn = args.output - add_file(args.input) + if (args.output): + destn = args.output + add_file(args.input) finally: - if (not destn): - destn.close() + if (not destn): + destn.close() diff --git a/build/single_file_libs/combine.sh b/build/single_file_libs/combine.sh index 8eac4f9e..674a5a27 100755 --- a/build/single_file_libs/combine.sh +++ b/build/single_file_libs/combine.sh @@ -200,6 +200,7 @@ if [ -n "$1" ]; then printf "" > "$DESTN" fi test_deps + log_line "Processing using the slower shell script; this might take a while" add_file "$1" else echo "Input file not found: \"$1\"" diff --git a/build/single_file_libs/create_single_file_decoder.sh b/build/single_file_libs/create_single_file_decoder.sh index 1c8841d1..3c0c577d 100755 --- a/build/single_file_libs/create_single_file_decoder.sh +++ b/build/single_file_libs/create_single_file_decoder.sh @@ -4,12 +4,12 @@ ZSTD_SRC_ROOT="../../lib" # Amalgamate the sources -echo "Amalgamating files... this can take a while" +echo "Amalgamating files..." # Using the faster Python script if we have 3.8 or higher if python3 -c 'import sys; assert sys.version_info >= (3,8)' 2>/dev/null; then - ./combine.py -r "$ZSTD_SRC_ROOT" -o zstddeclib.c zstddeclib-in.c + ./combine.py -r "$ZSTD_SRC_ROOT" -x legacy/zstd_legacy.h -o zstddeclib.c zstddeclib-in.c else - ./combine.sh -r "$ZSTD_SRC_ROOT" -o zstddeclib.c zstddeclib-in.c + ./combine.sh -r "$ZSTD_SRC_ROOT" -x legacy/zstd_legacy.h -o zstddeclib.c zstddeclib-in.c fi # Did combining work? if [ $? -ne 0 ]; then diff --git a/build/single_file_libs/create_single_file_library.sh b/build/single_file_libs/create_single_file_library.sh index 9b2f22a9..a6f71f0f 100755 --- a/build/single_file_libs/create_single_file_library.sh +++ b/build/single_file_libs/create_single_file_library.sh @@ -4,12 +4,12 @@ ZSTD_SRC_ROOT="../../lib" # Amalgamate the sources -echo "Amalgamating files... this can take a while" +echo "Amalgamating files..." # Using the faster Python script if we have 3.8 or higher if python3 -c 'import sys; assert sys.version_info >= (3,8)' 2>/dev/null; then - ./combine.py -r "$ZSTD_SRC_ROOT" -o zstd.c zstd-in.c + ./combine.py -r "$ZSTD_SRC_ROOT" -x legacy/zstd_legacy.h -o zstd.c zstd-in.c else - ./combine.sh -r "$ZSTD_SRC_ROOT" -o zstd.c zstd-in.c + ./combine.sh -r "$ZSTD_SRC_ROOT" -x legacy/zstd_legacy.h -o zstd.c zstd-in.c fi # Did combining work? if [ $? -ne 0 ]; then diff --git a/build/single_file_libs/zstd-in.c b/build/single_file_libs/zstd-in.c index b6946813..c22e4cd9 100644 --- a/build/single_file_libs/zstd-in.c +++ b/build/single_file_libs/zstd-in.c @@ -4,7 +4,7 @@ * * Generate using: * \code - * combine.sh -r ../../lib -o zstd.c zstd-in.c + * combine.sh -r ../../lib -x legacy/zstd_legacy.h -o zstd.c zstd-in.c * \endcode */ /* diff --git a/build/single_file_libs/zstddeclib-in.c b/build/single_file_libs/zstddeclib-in.c index 72abe613..42c51073 100644 --- a/build/single_file_libs/zstddeclib-in.c +++ b/build/single_file_libs/zstddeclib-in.c @@ -4,7 +4,7 @@ * * Generate using: * \code - * combine.sh -r ../../lib -o zstddeclib.c zstddeclib-in.c + * combine.sh -r ../../lib -x legacy/zstd_legacy.h -o zstddeclib.c zstddeclib-in.c * \endcode */ /*