tsv2json: first normalization algorithms working

This commit is contained in:
Robert Bendun 2023-03-18 20:24:20 +01:00
parent 4b89e6d350
commit 7c3a79b7f7
6 changed files with 360 additions and 12 deletions

3
.gitignore vendored
View File

@ -3,3 +3,6 @@ columns.pruned.tsv
columns.original.tsv
*.csv
csv2tsv/csv2tsv
.cache
compile_commands.json
tsv2json/tsv2json

3
.gitmodules vendored
View File

@ -1,3 +1,6 @@
[submodule "aitech-ium"]
path = aitech-ium
url = https://git.wmi.amu.edu.pl/AITech/aitech-ium.git
[submodule "tsv2json/json"]
path = tsv2json/json
url = https://github.com/RobertBendun/ImmidiateJSON.git

View File

@ -30,4 +30,7 @@ clean:
csv2tsv/csv2tsv: csv2tsv/csv2tsv.go
cd csv2tsv; go build
tsv2json/tsv2json: tsv2json/tsv2json.cc
g++ -std=c++20 -O3 -Wall -Wextra -o $@ $<
.PHONY: clean

View File

@ -2,7 +2,7 @@ Keep Type Index Column Description
y lower 1 title The official title of the language
y int 2 appeared What year was the language publicly released and/or announced?
y lower 3 type Which category in PLDB's subjective ontology does this entity fit into.
y int 4 pldbId computed
y lower 4 pldbId computed
y int 5 rank computed
y int 6 languageRank computed
y int 7 factCount computed
@ -13,12 +13,12 @@ y int 11 paperCount computed
y int 12 numberOfUsers computed
y int 13 numberOfJobs computed
y int 14 githubBigQuery.repos How many repos for this language are listed in Google's BigQuery Public GitHub Dataset snapshot.
y lower(sep("\band\b")) 15 creators Name(s) of the original creators of the language delimited by " and "
y sep(" and ").lower 15 creators Name(s) of the original creators of the language delimited by " and "
y bool 16 githubRepo URL of the official GitHub repo for the project if it hosted there.
y bool 17 website URL of the official homepage for the language project.
y bool 18 wikipedia URL of the entity on Wikipedia, if and only if it has a page dedicated to it.
y lower(sep("\b&&\b")) 19 originCommunity In what community(ies) did the language first originate?
y lower(skip("various",sep("\band\b"))) 20 country What country was the language first developed in?
y sep(" && ").lower 19 originCommunity In what community(ies) did the language first originate?
y unless("various").lower.sep(" and ") 20 country What country was the language first developed in?
n Type 21 centralPackageRepositoryCount If you've searched for a CPM for this language and can't find one, set 0 as the count.
n Type 22 reference A link to more info about this entity. You can add raw links and then auto "upgrade" them using some of the importer code.
n Type 23 hopl The matching language on Diarmuid Pigott's Online Historical Encyclopaedia of Programming Languages site (https://hopl.info/)
@ -85,9 +85,9 @@ y bool 83 features.hasMultiLineComments
n Type 84 rijuRepl A link to try this language on riju.codes
n Type 85 githubLanguage.codemirror_mime_type A String name of the file mime type used for highlighting whenever a file is edited. This should match the `mime` associated with the mode from https://git.io/f4SoQ
n Type 86 githubLanguage.codemirror_mode A String name of the CodeMirror Mode used for highlighting whenever a file is edited. This must match a mode from https://git.io/vi9Fx
y sep(' ') 87 fileExtensions An Array of associated extensions (the first one is considered the primary extension, the others should be listed alphabetically).
y sep(" ") 87 fileExtensions An Array of associated extensions (the first one is considered the primary extension, the others should be listed alphabetically).
n Type 88 tiobe Tiobe maintains a well known ranking of programming languages here: https://www.tiobe.com/tiobe-index/
y sep(' ') 89 related What languages are related? This serves as a catch all, and it is better to use a more specific relationship node such as "supersetOf".
y sep(" ") 89 related What languages are related? This serves as a catch all, and it is better to use a more specific relationship node such as "supersetOf".
y str 90 multiLineCommentTokens A comment with a start delimiter and end token (which can be the same) that can span multiple lines.
n Type 91 aka Another name for the language. Entries can have multiple aka lines.
y bool 92 features.hasIntegers
@ -95,7 +95,7 @@ n Type 93 helloWorldCollection Hello world written in this language from http://
n Type 94 githubLanguage.aliases An Array of additional aliases (implicitly includes name.downcase).
y bool 95 features.hasFloats
n Type 96 tryItOnline A link to try this language on https://tio.run
y sep(' ') 97 writtenIn What language(s) is the main implementation written in?
y sep(" ") 97 writtenIn What language(s) is the main implementation written in?
y bool 98 features.hasBooleans
n Type 99 keywords What are all the keywords in this language?
n Type 100 indeedJobs How many job descriptions match this query for this language on indeed.com?
@ -139,10 +139,10 @@ n Type 137 tiobe.currentRank What is the current Tiobe rank of this language?
y bool 138 features.hasWhileLoops
n Type 139 forLanguages Which languages is this repository for?
n Type 140 packageCount How many packages are in the repository? A package is some code with a name and a namespace, shipped as an atomic unit, with an owner(s).
y sep(' ') 141 supersetOf Is this language a superset of another? If you specify this link then the superset language will inherit all features of subset language.
y sep(" ") 141 supersetOf Is this language a superset of another? If you specify this link then the superset language will inherit all features of subset language.
n Type 142 indeedJobs.2017
y bool 143 features.hasBinaryNumbers
y sep(' ') 144 influencedBy What languages influenced this one?
y sep(" ") 144 influencedBy What languages influenced this one?
y bool 145 features.hasOperatorOverloading
y bool 146 features.hasImports
y bool 147 features.hasFunctions
@ -198,7 +198,7 @@ n Type 196 demoVideo Provide a url of a demo video of the language.
n Type 197 isPublicDomain Is it public domain?
y bool 198 features.hasMultilineStrings
y bool 199 features.hasVariableSubstitutionSyntax Do you use different syntax when assigning versus referencing a variable?
y sep(' ') 200 subsetOf Is this language a subset of another?
y sep(" ") 200 subsetOf Is this language a subset of another?
n Type 201 firstAnnouncement A url announcing the creation or release of a new language
n Type 202 packageInstallCount How many packages have been downloaded?
y bool 203 features.canWriteToDisk
@ -273,7 +273,7 @@ y bool 271 features.hasDestructuring
y bool 272 features.hasGenerators
y bool 273 features.hasDynamicProperties
y bool 274 features.hasExpressions
y sep(' ') 275 forkOf What language is this language a fork of?
y sep(" ") 275 forkOf What language is this language a fork of?
n Type 276 inputLanguages Which language(s) does this take as input? For compilers, what languages does this compile compile?
n Type 277 redditDiscussion A link to a related discussion on reddit.
y bool 278 features.hasTryCatch
@ -347,7 +347,7 @@ n Type 345 ebook Link to a free eBook about this. Only include if the eBook is o
y bool 346 features.hasExports
y bool 347 features.hasZippers
y bool 348 features.hasMonads
y sep(' ') 349 extensionOf What language is this language an extension of?
y sep(" ") 349 extensionOf What language is this language an extension of?
n Type 350 zulip Link to official (or popular unofficial) Zulip for language development.
y bool 351 features.hasImplicitArguments
y bool 352 features.hasDynamicTyping

Can't render this file because it contains an unexpected character in line 16 and column 7.

1
tsv2json/json Submodule

@ -0,0 +1 @@
Subproject commit 822784d0ebce101249e38d928ce69033e30455f4

338
tsv2json/tsv2json.cc Normal file
View File

@ -0,0 +1,338 @@
#define IMM_JSON_IMPLEMENTATION
#include "json/imm_json.hh"
#include <cassert>
#include <cctype>
#include <fstream>
#include <iomanip>
#include <iostream>
#include <memory>
#include <optional>
#include <span>
#include <utility>
namespace split
{
struct sentinel {};
struct iterator
{
using difference_type = ptrdiff_t;
using value_type = std::string_view;
using iterator_category = std::input_iterator_tag;
using pointer = void;
using reference = std::string_view&;
explicit iterator(std::convertible_to<std::string_view> auto&& source, char delim)
: source{source}
, delim{delim}
{
++*this; // Compute first cell
}
inline iterator begin() const
{
return *this;
}
inline sentinel end() const
{
return sentinel{};
}
inline bool operator==(sentinel) const
{
return reached_end;
}
inline iterator& operator++()
{
if (source.empty()) {
reached_end = true;
return *this;
}
if (auto tab = source.find(delim); tab != std::string_view::npos) {
current = source.substr(0, tab);
source.remove_prefix(tab+1);
} else {
current = source;
source = {};
}
return *this;
}
inline iterator operator++(int)
{
auto copy = *this;
++*this;
return copy;
}
inline std::string_view operator*() const
{
return current;
}
std::string_view current;
std::string_view source;
char delim;
bool reached_end = false;
};
}
struct Expression
{
enum Type
{
Symbol,
Call,
};
Type type;
std::string_view symbol;
std::vector<Expression> sub;
Expression(std::string_view symbol)
: type{Type::Symbol}
, symbol{symbol}
{
}
Expression(std::string_view name, Expression &&arg)
: type{Type::Call}
, symbol{name}
, sub{std::move(arg)}
{
}
Expression(Expression const&) = default;
Expression(Expression &&) = default;
friend std::ostream& operator<<(std::ostream& os, Expression const& expr)
{
switch (expr.type) {
break; case Type::Symbol: os << std::quoted(expr.symbol);
break; case Type::Call: os << expr.symbol << '(' << expr.sub.front() << ')';
}
return os;
}
};
std::vector<Expression> parse_normalization_expression(std::string_view &source)
{
auto const skip_ws = [&] {
if (auto after_ws = source.find_first_not_of(" \t"); after_ws != std::string_view::npos) {
source.remove_prefix(after_ws);
}
};
std::vector<Expression> sequence;
for (;;) {
std::string_view symbol;
skip_ws();
// FIXME String escaping
if (source.starts_with('"')) {
source.remove_prefix(1);
auto const string_end = source.find('"');
if (string_end == std::string_view::npos) {
std::cerr << "[ERROR] Failed to parse '" << source << "': expected end of string\n";
std::exit(2);
}
std::string_view symbol{source.begin(), string_end};
source.remove_prefix(string_end+1);
skip_ws();
sequence.emplace_back(symbol);
goto next;
}
{
// Don't use islower since it uses locale (slow)
auto const symbol_end = std::find_if_not(source.begin(), source.end(), [](char c) { return c >= 'a' && c <= 'z'; });
if (symbol_end == source.begin()) {
std::cerr << "[ERROR] Failed to parse '" << source << "': expected symbol\n";
std::exit(2);
}
symbol = std::string_view{source.begin(), symbol_end};
source.remove_prefix(symbol.size());
skip_ws();
}
if (source.empty()) {
sequence.emplace_back(symbol);
goto next;
}
if (source.starts_with("(")) {
source.remove_prefix(1);
// FIXME Should separate expression sequence and expression argument parsing
sequence.emplace_back(symbol, std::move(parse_normalization_expression(source).front()));
skip_ws();
if (!source.starts_with(")")) {
std::cerr << "[ERROR] Failed to parse '" << source << "': expected closing bracket\n";
std::exit(2);
}
source.remove_prefix(1);
goto next;
}
next:
skip_ws();
if (source.starts_with('.')) {
source.remove_prefix(1);
continue;
} else {
break;
}
}
return sequence;
}
struct Value
{
enum class Type
{
String,
Number,
};
Type type;
std::string string;
double number;
explicit Value(std::string_view s)
: type(Type::String)
, string(s)
{
}
explicit Value(double number)
: type(Type::Number)
, number(number)
{
}
};
using Builtin = Value(*)(Value, std::optional<std::string_view>);
using Env = std::unordered_map<std::string, Builtin>;
Value eval(std::span<Expression> expressions, Value value, Env const& env)
{
for (auto const& expr : expressions) {
auto builtin = env.find(std::string(expr.symbol));
if (builtin == env.end()) {
std::cerr << "[ERROR] Unknown builtin: " << expr.symbol << '\n';
std::exit(1);
}
switch (expr.type) {
break; case Expression::Symbol:
value = builtin->second(std::move(value), std::nullopt);
break; case Expression::Call:
assert(expr.sub.size() == 1);
assert(expr.sub.front().type == Expression::Symbol);
value = builtin->second(std::move(value), expr.sub[0].symbol);
}
}
return value;
}
struct Column
{
Column(std::string_view name, std::string_view normalization_expression)
: name(name)
, expression_source(normalization_expression)
, expression(parse_normalization_expression(normalization_expression))
{
}
inline Value normalize(std::string_view source, Env const& env)
{
return eval(std::span(expression), Value(source), env);
}
std::string_view name;
std::string_view expression_source;
std::vector<Expression> expression;
};
int main(int argc, char** argv)
{
if (argc != 2) {
std::cerr << "usage: " << argv[0] << " <columns.tsv>\n";
std::cerr << " convert tsv file from TSV using definitions from columns.tsv\n";
}
std::ifstream columns_file(argv[1]);
static std::string source{std::istreambuf_iterator<char>(columns_file), {}};
if (auto it = std::next(split::iterator(source, '\t'), 1); it == split::sentinel{} || *it != "Type") {
std::cerr << "[ERROR] Expected Type description in column 2\n";
return 1;
}
std::vector<Column> columns;
for (std::string_view line : split::iterator(source, '\n')) {
auto tsv_it = split::iterator(line, '\t');
if (tsv_it == split::sentinel{} || *tsv_it++ != "y") {
continue;
}
auto const type = *tsv_it++; if (tsv_it == split::sentinel{}) continue;
[[maybe_unused]] auto const _column_number = *tsv_it++; if (tsv_it == split::sentinel{}) continue;
auto const name = *tsv_it++; if (tsv_it == split::sentinel{}) continue;
columns.emplace_back(name, type);
}
Env env = {
std::pair<std::string, Builtin> { "lower", +[](Value v, std::optional<std::string_view>) -> Value {
assert(v.type == Value::Type::String);
// FIXME Proper UTF-8 lowercase
// However, manual inspection of used TSV files prooved that there aren't any non-ascii uppercase letters
for (char &c : v.string) {
if (c >= 'A' && c <= 'Z') {
c = c - 'A' + 'a';
}
}
return v;
}},
std::pair<std::string, Builtin> { "int", +[](Value v, std::optional<std::string_view>) -> Value {
assert(v.type == Value::Type::String);
long long int n;
std::cout.flush();
auto [p, ec] = std::from_chars(v.string.data(), v.string.data() + v.string.size(), n);
assert(ec == std::errc{});
return Value(double(n));
}},
};
{
bool passed_header = false;
Json json;
auto _array = json.array();
for (std::string line; std::getline(std::cin, line); ) {
if (!passed_header) {
passed_header = true;
continue;
}
auto _object = json.object();
auto tsv = split::iterator(line, '\t');
for (auto i = 0u; tsv != split::sentinel{}; ++i, ++tsv) {
auto normalized = columns[i].normalize(*tsv, env);
switch (normalized.type) {
break; case Value::Type::String: json.key(columns[i].name) = normalized.string;
break; case Value::Type::Number: json.key(columns[i].name) = double(normalized.number);
}
}
break;
}
}
return 0;
}