From b971e406dee82486d53737a928b0bb3b482a6c49 Mon Sep 17 00:00:00 2001 From: Eelco Dolstra Date: Fri, 31 May 2019 19:01:11 +0200 Subject: [PATCH] Support 'dir' and other parameters in path flakerefs --- src/libexpr/primops/flakeref.cc | 90 ++++++++++++++++++--------------- 1 file changed, 49 insertions(+), 41 deletions(-) diff --git a/src/libexpr/primops/flakeref.cc b/src/libexpr/primops/flakeref.cc index 3c805eff8..306da20fe 100644 --- a/src/libexpr/primops/flakeref.cc +++ b/src/libexpr/primops/flakeref.cc @@ -1,4 +1,5 @@ #include "flakeref.hh" +#include "store-api.hh" #include @@ -30,10 +31,6 @@ const static std::string schemeRegex = "(?:http|https|ssh|git|file)"; const static std::string authorityRegex = "[a-zA-Z0-9._~-]*"; const static std::string segmentRegex = "[a-zA-Z0-9._~-]+"; const static std::string pathRegex = "/?" + segmentRegex + "(?:/" + segmentRegex + ")*"; -// FIXME: support escaping in query string. -// Note: '/' is not a valid query parameter, but so what... -const static std::string paramRegex = "[a-z]+=[/a-zA-Z0-9._-]*"; -const static std::string paramsRegex = "(?:[?](" + paramRegex + "(?:&" + paramRegex + ")*))"; // 'dir' path elements cannot start with a '.'. We also reject // potentially dangerous characters like ';'. @@ -41,7 +38,7 @@ const static std::string subDirElemRegex = "(?:[a-zA-Z0-9_-]+[a-zA-Z0-9._-]*)"; const static std::string subDirRegex = subDirElemRegex + "(?:/" + subDirElemRegex + ")*"; -FlakeRef::FlakeRef(const std::string & uri, bool allowRelative) +FlakeRef::FlakeRef(const std::string & uri_, bool allowRelative) { // FIXME: could combine this into one regex. @@ -50,21 +47,46 @@ FlakeRef::FlakeRef(const std::string & uri, bool allowRelative) std::regex::ECMAScript); static std::regex githubRegex( - "github:(" + ownerRegex + ")/(" + repoRegex + ")(?:/" + revOrRefRegex + ")?" - + paramsRegex + "?", + "github:(" + ownerRegex + ")/(" + repoRegex + ")(?:/" + revOrRefRegex + ")?", std::regex::ECMAScript); static std::regex uriRegex( "((" + schemeRegex + "):" + "(?://(" + authorityRegex + "))?" + - "(" + pathRegex + "))" + - paramsRegex + "?", + "(" + pathRegex + "))", std::regex::ECMAScript); static std::regex refRegex2(refRegex, std::regex::ECMAScript); static std::regex subDirRegex2(subDirRegex, std::regex::ECMAScript); + auto [uri, params] = splitUriAndParams(uri_); + + auto handleSubdir = [&](const std::string & name, const std::string & value) { + if (name == "dir") { + if (value != "" && !std::regex_match(value, subDirRegex2)) + throw Error("flake '%s' has invalid subdirectory '%s'", uri, value); + subdir = value; + return true; + } else + return false; + }; + + auto handleGitParams = [&](const std::string & name, const std::string & value) { + if (name == "rev") { + if (!std::regex_match(value, revRegex)) + throw Error("invalid Git revision '%s'", value); + rev = Hash(value, htSHA1); + } else if (name == "ref") { + if (!std::regex_match(value, refRegex2)) + throw Error("invalid Git ref '%s'", value); + ref = value; + } else if (handleSubdir(name, value)) + ; + else return false; + return true; + }; + std::cmatch match; if (std::regex_match(uri.c_str(), match, flakeRegex)) { IsAlias d; @@ -88,17 +110,11 @@ FlakeRef::FlakeRef(const std::string & uri, bool allowRelative) else if (match[4].matched) { ref = match[4]; } - for (auto & param : tokenizeString(match[5], "&")) { - auto n = param.find('='); - assert(n != param.npos); - std::string name(param, 0, n); - std::string value(param, n + 1); - if (name == "dir") { - if (value != "" && !std::regex_match(value, subDirRegex2)) - throw Error("flake '%s' has invalid subdirectory '%s'", uri, value); - subdir = value; - } else - throw Error("invalid Git flake reference parameter '%s', in '%s'", name, uri); + for (auto & param : params) { + if (handleSubdir(param.first, param.second)) + ; + else + throw Error("invalid Git flakeref parameter '%s', in '%s'", param.first, uri); } data = d; } @@ -108,26 +124,12 @@ FlakeRef::FlakeRef(const std::string & uri, bool allowRelative) { IsGit d; d.uri = match[1]; - for (auto & param : tokenizeString(match[5], "&")) { - auto n = param.find('='); - assert(n != param.npos); - std::string name(param, 0, n); - std::string value(param, n + 1); - if (name == "rev") { - if (!std::regex_match(value, revRegex)) - throw Error("invalid Git revision '%s'", value); - rev = Hash(value, htSHA1); - } else if (name == "ref") { - if (!std::regex_match(value, refRegex2)) - throw Error("invalid Git ref '%s'", value); - ref = value; - } else if (name == "dir") { - if (value != "" && !std::regex_match(value, subDirRegex2)) - throw Error("flake '%s' has invalid subdirectory '%s'", uri, value); - subdir = value; - } else + for (auto & param : params) { + if (handleGitParams(param.first, param.second)) + ; + else // FIXME: should probably pass through unknown parameters - throw Error("invalid Git flake reference parameter '%s', in '%s'", name, uri); + throw Error("invalid Git flakeref parameter '%s', in '%s'", param.first, uri); } if (rev && !ref) throw Error("flake URI '%s' lacks a Git ref", uri); @@ -138,6 +140,12 @@ FlakeRef::FlakeRef(const std::string & uri, bool allowRelative) IsPath d; d.path = allowRelative ? absPath(uri) : canonPath(uri); data = d; + for (auto & param : params) { + if (handleGitParams(param.first, param.second)) + ; + else + throw Error("invalid Git flakeref parameter '%s', in '%s'", param.first, uri); + } } else @@ -165,10 +173,10 @@ std::string FlakeRef::to_string() const } else if (auto refData = std::get_if(&data)) { - assert(subdir == ""); + string = refData->path; if (ref) addParam("ref", *ref); if (rev) addParam("rev", rev->gitRev()); - return refData->path; + if (subdir != "") addParam("dir", subdir); } else if (auto refData = std::get_if(&data)) {