_pakfire: Remove the trailing newline before passing to python

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

execute: Pass string length to logging function

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

execute: Log something if the logging callback failed

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

_pakfire: Translate log levels to Python

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

parser: Drop unused function

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

parser: Consolidate compiling regular expressions into one function

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

parser: commands: Don't be greedy when finding the closing bracket

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

builder: Refactor logging

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

parser: Capture the output of commands

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

util: Make pakfire_string_replace more robust on invalid inputs

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

execute: Add data pointer to logging callback

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

parser: Implement executing commands in %(...)

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

parser: Use JIT for PCRE2

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

parser: Split expanding variables into separate function

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

parser: Use PCRE2 for regex matching

The regular expression is only compiled once per parser when it is being
needed and not every time we are searching for variables.

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

libpakfire: Link against PCRE2 (8 bit)

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

parser: expand: Tidy up some code

No functional changes

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

tests: makefile: Update test to latest parser changes

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

parser: Improve performance of expanding strings

By using pakfire_string_replace, all occurrences of a variable will be
replaced at once until no variables are being left.

It can still happen, that the same variable is being matched later again
(multiple times even), but it looks to me like adding any caching
mechanism adds a lot more overhead.

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

parser: Add support for flags

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

util: Implement a function that replace patterns in a string

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

builder: Set BASEDIR in build environment

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

python: parser: Implement setting variables

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

builder: Run build scripts in all stages

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

python: Export pakfire_execute_script

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

execute: Activate environment earlier to write to ramdisks

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

parser: Split keys into name and namespace part

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

tests: Check and warn if an expected string is NULL

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

execute: Add function to execute a script

This has been moved from step

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

_pakfire: parser: Hold a reference to the parser object

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

_pakfire: archive: Remove debugging line

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

parser: Handle appending variables correctly

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

parser: Handle "export" keyword

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

parser: Fix reference counting

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

parser: Unref parent only when set

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

tests: Return number of failed test

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

parser: Have the merge() function return an integer

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

parser: Remove unused function

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

builder: Implement reading makefiles

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

parser: Create an error object when parsing failed

This can be used to report more details about the parsing error to the
user

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

parser: Handle lines continued with backslash

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

tests: makefile: Test if parsing macros works

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

Fix typo

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

_pakfire: Show path when mount/umount failed

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

parser: Allow empty lines after T_KEY to open a block

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

tests: Parse beep.nm

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

util: Refactor pakfire_string_partition

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

parser: Disable debugging

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

parser: Do not consider a READLINES block opening with fewer indents

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

parser: Allow reading empty keys

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

parser: Remove hack to match beginning of line

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

parser: Fixup readline

This could read into the next line when a line was empty

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

parser: Just ignore any "def" or "define" keywords

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

parser: Add a hack for "template NAME" lines

These will now be parsed as empty subparsers which are being turned into
a regular declaration.

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

parser: Remove hack to detect "end"

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

parser: Add support for named subparsers and variables

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

parser: Only treat keywords as such when they are at the beginning of the line

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

parser: Handle empty else statement returning NULL

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

tests: parser: Add a simple if clause

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

parser: Remove any leading whitespace from values

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

parser: Change handling of namespace

The namespace is being added when parsers are being merged

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

tests: Show received string as well as expected string

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

parser: Drop unused function

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

parser: Introduce new data type for declarations

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

parser: Add support for conditionals

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

parser: Do not define yyunput

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

parser: Remove some debugging output

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

parser: Allow empty lines in READLINES

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

parser: Add keywords that open a new parser

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

parser: Jump into an extra context when we could not find a keyword

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

parser: Use yyless(0) instead of unput(*yytext)

unput fails for the first character in the body that is being scanned
and yyless is more efficient than unput, too.

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

parser: Move test cases into many smaller files

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

parser: Ignore empty lines

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

parser: Parse multi-line definitions

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

parser: Implement appending strings with +=

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

parser: Start from scratch

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

repo: Increase XZ work buffer from 1 byte to 1 megabyte

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

parser: Remove more useless directives

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

parser: Move cleaning up the parser out of debug block

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

parser: Revert back to GLR mode

We need to look ahead for more than one token which requires us using
GLR.

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

parser: Remove ineffective directives

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

configure: Fix raw bison and call it directly

The old YACC macro has some issues with using modern bison features and
always enables compatibility mode with yacc.

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

parser: Fix path of source files after move

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

configure: Use bison instead of POSIX yacc

bison is around everywhere and we can make use of its modern features
instead of sticking to the POSIX standard.

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

python: Export pakfire_make_path()

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

parser: Disable useless code when compiling without debug

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

parser: Do not use pakfire's own memory allocation functions any more

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

bison: Update %error-verbose directive

This is deprecated and has been replaced

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

python: parser: Extend and let it read from file

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

python: Add simply Parser module

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

pakfire: Skip deactivation when running in /

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

parser: Remove declaration limit and allocate memory dynamically

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

builder: Extract packages into the build environment

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

Drop unused compression module

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

system: Drop loads of unused mountpoint code

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

_pakfire: Drop unused sync() function

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

_pakfire: Drop unused constants

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

builder: Add mount/umount to _pakfire to avoid calling /bin/mount

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

builder: Remove ccache settings in environment

This is something that we should do elsewhere

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>

builder: Do not lock build environments

This feels a bit unnecessary because the paths are being randomised
anyways and this might need to move into Pakfire().

Signed-off-by: Michael Tremer <michael.tremer@ipfire.org>