Improved dependency search on rpmfind.net: we now include the distro

code (i.e., fc26, fc27, etc) and have a placeholder to check versions
when a range filter is given. We can probably use some of the code from
Scripts' FindDependencies.c here.

4 files changed, 101 insertions, 30 deletions

diff --git a/Functions/DEB b/Functions/DEB
index 1c0e498..5d59135 100644
--- a/Functions/DEB
+++ b/Functions/DEB
@@ -17,7 +17,7 @@ function thirdparty_arch() {
     fi
 }
 
-function thirdparty_distribution() {
+function thirdparty_distribution_name() {
     local debfile="$1"
     local distro=$(dpkg-deb --field "$debfile" "Distribution")
     if [ -z "$distro" ]
@@ -26,6 +26,10 @@ function thirdparty_distribution() {
     fi
 }
 
+function thirdparty_distribution_code() {
+    return
+}
+
 function thirdparty_dependencies() {
     local debfile="$1"
     local deps=$(dpkg-deb --field "$debfile" "Depends")
diff --git a/Functions/RPM b/Functions/RPM
index 635f643..a166fb0 100644
--- a/Functions/RPM
+++ b/Functions/RPM
@@ -11,11 +11,16 @@ function thirdparty_arch() {
     rpminfo --arch "$rpmfile"
 }
 
-function thirdparty_distribution() {
+function thirdparty_distribution_name() {
     local rpmfile="$1"
     rpminfo --distribution "$rpmfile"
 }
 
+function thirdparty_distribution_code() {
+    local rpmfile="$1"
+    echo "$rpmfile" | awk -F. '{print $(NF-2)}'
+}
+
 function thirdparty_dependencies() {
     local rpmfile="$1"
     rpminfo --dependencies "$rpmfile"
@@ -64,8 +69,12 @@ function thirdparty_url() {
 function thirdparty_search_remotedb() {
     local path="$1"
     local arch="$2"
-    local distro="$3"
-    RPMFinder --path="$path" --arch="$arch" --distro="$distro"
+    local distroname="$3"
+    local distrocode="$4"
+    local result=$(RPMFinder --path="$path" --arch="$arch" --distroname="$distroname" --distrocode="$distrocode")
+    [ -z "$result" ] && [ "$arch" = "noarch" ] && result=$(RPMFinder --path="$path" --arch="$(uname -m)" --distroname="$distroname" --distrocode="$distrocode")
+    [ -z "$result" ] && return
+    echo "$result"
 }
 
 function thirdparty_uncompress() {
diff --git a/bin/RPMFinder b/bin/RPMFinder
index 995bb8a..1b8bf2f 100755
--- a/bin/RPMFinder
+++ b/bin/RPMFinder
@@ -9,6 +9,7 @@
 # Released under the GNU GPL version 2 or above.
 
 import os
+import re
 import sys
 import argparse
 import subprocess
@@ -34,53 +35,98 @@ class RPMFind_Parser(HTMLParser):
 		if len(self.tags) and self.tags[-1] == "a" and data.find(".rpm") >= 0:
 			self.candidates.append(data)
 
-	def get_pkgname(self):
+	def get_pkgnames(self):
 		if len(self.candidates) == 0:
 			return ""
-		name = os.path.commonprefix(self.candidates)
-		if name.endswith("-"):
-			name = name[:-1]
-		return name
+		return self.candidates
 
 
 class RPMFinder:
-	def find(self, path, arch, distro):
+	def find(self, path, arch, distroname, distrocode):
 		'''
 		Searches rpmfind.net for a given file. Arch and distro can
-		be provided to reduce the search scope. Returns the package
+		be provided to narrow the search scope. Returns the package
 		name on success or an empty string if no matches were found.
 		'''
 		self.path = path
 		self.arch = arch
-		self.distro = distro
-		return self.__search_rpmfind_net()
+		self.distroname = distroname.replace(" ", "+")
+		self.distrocode = distrocode
+		self.name = path
+		for token in [">=", ">", "<=", "<", "="]:
+			self.name = self.name.replace(token, " ")
+		self.name = self.name.split(" ")[0]
+
+		matches = self.__search_rpmfind_net()
+		if len(matches) == 0:
+			return []
+		if any(op in self.path for op in [">", "<", "="]):
+			op, version = self.__get_op_and_version()
+			return self.__filter(matches, op, version)
+		else:
+			return matches
+
+	def __filter(self, matches, op, version):
+		filtered = []
+		for match in matches:
+			pkg_version = match.replace(self.name, "").replace("{0}.{1}.rpm".format(self.distrocode, self.arch), "").strip("-").strip(".")
+			sys.stderr.write("package {0}: {1} {2} {3}?\n".format(match, pkg_version, op, version))
+
+			# TODO: filter based on op and version
+			if len(filtered) == 0:
+				filtered.append(match)
+		return filtered
+
+	def __get_op_and_version(self):
+		if ">=" in self.path:
+			return ">=", self.path.split(">=")[1].strip()
+		elif ">" in self.path:
+			return ">", self.path.split(">")[1].strip()
+		elif "<=" in self.path:
+			return "<=", self.path.split("<=")[1].strip()
+		elif "<" in self.path:
+			return "<", self.path.split("<")[1].strip()
+		elif "=" in self.path:
+			return "=", self.path.split("=")[1].strip()
+		else:
+			sys.stderr.write("could not extract op and version from {}\n".format(self.path))
+			return None, None
 
 	def __search_rpmfind_net(self):
 		path = self.path.replace("/", "%2F")
-		arch = self.arch
 		baseuri = "http://rpmfind.net/linux/rpm2html/search.php"
-		query = "?query={0}&submit=Search+...&system=&arch={1}".format(path, arch)
+		query = "?query={0}&submit=Search+...&system={1}&arch={2}".format(path, self.distroname, self.arch)
 		html = subprocess.check_output(["wget", "--quiet", "{0}{1}".format(baseuri, query), "-O", "-"])
 
+		# Compile a regex that catches package names derived from the basename given by self.path.
+		# Example: perl-DBICx when perl-DBI is wanted.
+		regex = re.compile(r"{0}\-[0-9]+.*{1}.{2}.rpm".format(self.name, self.distrocode, self.arch))
+
 		htmlparser = RPMFind_Parser()
 		htmlparser.feed(str(html))
-		return htmlparser.get_pkgname()
+		pkgnames = htmlparser.get_pkgnames()
+		result = []
+		for pkgname in pkgnames:
+			if regex.match(pkgname):
+				result.append(pkgname)
+		return result
 
 
 def main():
 	argparser = argparse.ArgumentParser(argument_default="")
 	argparser.add_argument("--path", type=str, help="File name to search for in the remote RPM databases")
 	argparser.add_argument("--arch", type=str, help="Architecture (optional)")
-	argparser.add_argument("--distro", type=str, help="Distribution (optional)")
+	argparser.add_argument("--distroname", type=str, help="Distribution name (optional)")
+	argparser.add_argument("--distrocode", type=str, help="Distribution code (optional)")
 	args = argparser.parse_args()
 
 	if len(args.path) == 0:
 		argparser.print_help()
 		sys.exit(1)
 
-	pkgname = RPMFinder().find(args.path, args.arch, args.distro)
+	pkgname = RPMFinder().find(args.path, args.arch, args.distroname, args.distrocode)
 	if len(pkgname):
-		print(pkgname)
+		print(pkgname[0])
 
 if __name__ == "__main__":
 	main()
diff --git a/bin/ThirdPartyInstaller b/bin/ThirdPartyInstaller
index caaf8b9..76d1755 100755
--- a/bin/ThirdPartyInstaller
+++ b/bin/ThirdPartyInstaller
@@ -9,7 +9,7 @@ Import OptionParser
 ### Options ###################################################################
 
 scriptDescription="Install RPM and DEB packages on GoboLinux."
-scriptCredits="Copyright (C) Lucas C. Villa Real, 2016,2017 - Released under the GNU GPL."
+scriptCredits="Copyright (C) Lucas C. Villa Real, 2016-2018 - Released under the GNU GPL."
 helpOnNoArguments=yes
 scriptUsage="<file.rpm> [file.rpm]"
 scriptExample="xispita-2.0.3-1.x86_64.rpm"
@@ -223,7 +223,7 @@ function populate_dependencies() {
         then
             Log_Verbose "Skipping dependency passed as input file: $dependency"
         else
-            depinfo=$(lookup_pkgname "$dependency")
+            depinfo=$(lookup_pkgname "$inputfile" "$dependency")
             if [ "$depinfo" ]
             then echo "$depinfo"
             else echo "# Unresolved dependency: $dependency"
@@ -237,7 +237,7 @@ function populate_resources() {
     local arch=$(thirdparty_arch "$inputfile")
     local description=$(thirdparty_description "$inputfile")
     local release=$(thirdparty_release "$inputfile")
-    local distro=$(thirdparty_distribution "$inputfile")
+    local distro=$(thirdparty_distribution_name "$inputfile")
     
     Log_Normal "Populating Resources."
     mkdir -p Resources
@@ -307,7 +307,8 @@ function take_dependency_from_path() {
     local symbol="$3"
     local fullpath="$(readlink -f $path)"
     local arch=$(thirdparty_arch "$inputfile")
-    local distro=$(thirdparty_distribution "$inputfile")
+    local distro=$(thirdparty_distribution_name "$inputfile")
+    local distrocode=$(thirdparty_distribution_code "$inputfile")
 
     local depname=
     local depversion=
@@ -364,7 +365,7 @@ function take_dependency_from_path() {
         if Boolean "web"
         then
             Log_Normal "Searching the remote $(thirdparty_backend) database for the package hosting $originalpath"
-            depname=$(thirdparty_search_remotedb "$originalpath" "$arch" "$distro")
+            depname=$(thirdparty_search_remotedb "$originalpath" "$arch" "$distro" "$distrocode")
             if [ "$depname" ]
             then
                 # TODO: we could now lookup the GoboLinux recipe store to find whether we
@@ -377,7 +378,8 @@ function take_dependency_from_path() {
 }
 
 function lookup_pkgname() {
-    local dependency="$1"
+    local inputfile="$1"
+    local dependency="$2"
     local pkgname=$(echo "$dependency" | cut -d'(' -f1)
 
     # GoboLinux doesn't have "devel" packages like most mainstream distros do
@@ -393,20 +395,30 @@ function lookup_pkgname() {
         fi
     done
 
-    # Query the GoboLinux recipe store.
     if Boolean "web"
     then
+        # Query the GoboLinux recipe store.
         local recipeurl=$(FindPackage -t recipe "${pkgname}" || FindPackage -t recipe "${goboname}")
         if [ "$recipeurl" ]
         then
             # TODO we're potentially discarding the wanted version(s) of the given dep
             echo "$(basename $recipeurl | sed 's,\(.*\)--.*--.*,\1,g')" && return 0
         fi
-    fi
-
-    # TODO Query the remote RPM database for $pkgname
 
-    # TODO Query the remote DEB database for $pkgname
+        # Query the remote RPM database for $pkgname
+        local arch=$(thirdparty_arch "$inputfile")
+        local distro=$(thirdparty_distribution_name "$inputfile")
+        local distrocode=$(thirdparty_distribution_code "$inputfile")
+        Log_Normal "Searching the remote $(thirdparty_backend) database for the package hosting $pkgname"
+        local depname=$(thirdparty_search_remotedb "$pkgname" "$arch" "$distro" "$distrocode")
+        if [ "$depname" ]
+        then
+            # TODO: download and install the RPM
+            echo "$depname" && return 0
+        fi
+    
+        # TODO Query the remote DEB database for $pkgname
+    fi
 
     return 1
 }