From 80bf38904a525674d34222454de2fc2fc96056f2 Mon Sep 17 00:00:00 2001
From: Joey Riches <josephriches@gmail.com>
Date: Sat, 22 Feb 2025 14:41:16 +0000
Subject: [PATCH 1/2] boulder/draft: Allow transforming the given URL

We may want to sanitize the URL given in some cases.
---
 boulder/src/draft/metadata.rs        | 11 ++++++++++-
 boulder/src/draft/metadata/basic.rs  |  1 +
 boulder/src/draft/metadata/github.rs |  1 +
 3 files changed, 12 insertions(+), 1 deletion(-)

diff --git a/boulder/src/draft/metadata.rs b/boulder/src/draft/metadata.rs
index c775a06f..2e8585e0 100644
--- a/boulder/src/draft/metadata.rs
+++ b/boulder/src/draft/metadata.rs
@@ -20,6 +20,7 @@ pub struct Source {
     pub name: String,
     pub version: String,
     pub homepage: String,
+    pub uri: String,
 }
 
 impl Metadata {
@@ -45,7 +46,15 @@ impl Metadata {
     pub fn upstreams(&self) -> String {
         self.upstreams
             .iter()
-            .map(|Upstream { uri, hash }| format!("    - {uri} : {hash}"))
+            .enumerate()
+            .map(|(i, Upstream { uri, hash })| {
+                let uri_to_use = if i == 0 && !self.source.uri.is_empty() {
+                    &self.source.uri
+                } else {
+                    uri.as_str()
+                };
+                format!("    - {uri_to_use} : {hash}")
+            })
             .join("\n")
     }
 }
diff --git a/boulder/src/draft/metadata/basic.rs b/boulder/src/draft/metadata/basic.rs
index 43957956..40a0d89c 100644
--- a/boulder/src/draft/metadata/basic.rs
+++ b/boulder/src/draft/metadata/basic.rs
@@ -24,5 +24,6 @@ pub fn source(upstream: &Url) -> Option<Source> {
         name,
         version,
         homepage: homepage.to_owned(),
+        uri: upstream.to_string(),
     })
 }
diff --git a/boulder/src/draft/metadata/github.rs b/boulder/src/draft/metadata/github.rs
index 5c435789..989b34af 100644
--- a/boulder/src/draft/metadata/github.rs
+++ b/boulder/src/draft/metadata/github.rs
@@ -30,6 +30,7 @@ pub fn source(upstream: &Url) -> Option<Source> {
             name: project.to_lowercase(),
             version,
             homepage: format!("https://github.com/{owner}/{project}"),
+            uri: upstream.to_string(),
         });
     }
 

From 4fbebd76084a6718e03156d65c3b1e098410dea5 Mon Sep 17 00:00:00 2001
From: Joey Riches <josephriches@gmail.com>
Date: Sat, 22 Feb 2025 14:42:13 +0000
Subject: [PATCH 2/2] boulder/draft: Add metadata matching for pypi.org

---
 boulder/src/draft/metadata.rs      |  5 +++-
 boulder/src/draft/metadata/pypi.rs | 37 ++++++++++++++++++++++++++++++
 2 files changed, 41 insertions(+), 1 deletion(-)
 create mode 100644 boulder/src/draft/metadata/pypi.rs

diff --git a/boulder/src/draft/metadata.rs b/boulder/src/draft/metadata.rs
index 2e8585e0..0faf31b8 100644
--- a/boulder/src/draft/metadata.rs
+++ b/boulder/src/draft/metadata.rs
@@ -8,6 +8,7 @@ use super::Upstream;
 
 mod basic;
 mod github;
+mod pypi;
 
 #[derive(Default)]
 pub struct Metadata {
@@ -33,6 +34,7 @@ impl Metadata {
                 if let Some(matched) = match matcher {
                     Matcher::Basic => basic::source(&upstream.uri),
                     Matcher::Github => github::source(&upstream.uri),
+                    Matcher::Pypi => pypi::source(&upstream.uri),
                 } {
                     source = matched;
                     break;
@@ -62,8 +64,9 @@ impl Metadata {
 enum Matcher {
     Basic,
     Github,
+    Pypi,
 }
 
 impl Matcher {
-    const ALL: &'static [Self] = &[Self::Github, Self::Basic];
+    const ALL: &'static [Self] = &[Self::Github, Self::Pypi, Self::Basic];
 }
diff --git a/boulder/src/draft/metadata/pypi.rs b/boulder/src/draft/metadata/pypi.rs
new file mode 100644
index 00000000..99d1b086
--- /dev/null
+++ b/boulder/src/draft/metadata/pypi.rs
@@ -0,0 +1,37 @@
+// SPDX-FileCopyrightText: Copyright © 2020-2025 Serpent OS Developers
+//
+// SPDX-License-Identifier: MPL-2.0
+
+use regex::Regex;
+use url::Url;
+
+use crate::util;
+
+use super::Source;
+
+pub fn source(upstream: &Url) -> Option<Source> {
+    let regex = Regex::new(
+        r"^https://files\.pythonhosted\.org/packages/[a-f0-9]{2}/[a-f0-9]{2}/[a-f0-9]+/([^/]+)-([\d.]+)\.tar\.gz$",
+    )
+    .ok()?;
+
+    let filename = util::uri_file_name(upstream);
+
+    let captures = regex.captures(upstream.as_str())?;
+
+    let mut name = captures.get(1)?.as_str().to_owned();
+    let version = captures.get(2)?.as_str().to_owned();
+
+    let first_char = &name.chars().next().unwrap_or_default();
+
+    if !name.starts_with("python-") {
+        name = format!("python-{name}");
+    }
+
+    Some(Source {
+        name: name.to_owned(),
+        version,
+        homepage: format!("https://pypi.org/project/{name}"),
+        uri: format!("https://files.pythonhosted.org/packages/source/{first_char}/{name}/{filename}"),
+    })
+}