From 4476d9d02ac7af21a7bc6ef92490491d308237f4 Mon Sep 17 00:00:00 2001
From: Karen Arutyunov <karen@codesynthesis.com>
Date: Wed, 29 Apr 2020 22:20:21 +0300
Subject: Add --pkg-proxy common option

---
 bpkg/archive.cxx          |   3 -
 bpkg/archive.hxx          |   2 -
 bpkg/auth.cxx             |   1 -
 bpkg/checksum.cxx         |   4 --
 bpkg/common.cli           |  22 +++++++-
 bpkg/diagnostics.cxx      |  29 ++++++++--
 bpkg/diagnostics.hxx      |  22 ++++++++
 bpkg/fetch-git.cxx        |   1 -
 bpkg/fetch-pkg.cxx        |  12 +++-
 bpkg/fetch.cxx            | 138 ++++++++++++++++++++++++++++++++++++++++------
 bpkg/fetch.hxx            |  15 +++--
 bpkg/manifest-utility.cxx |   1 -
 bpkg/pkg-build.cxx        |   3 +-
 bpkg/pkg-unpack.cxx       |   3 -
 bpkg/pkg-verify.cxx       |   1 -
 bpkg/satisfaction.cxx     |   3 -
 bpkg/types-parsers.cxx    |  22 ++++++++
 bpkg/types-parsers.hxx    |  10 ++++
 bpkg/types.hxx            |   7 +++
 bpkg/utility.cxx          |   1 -
 bpkg/utility.hxx          |   5 ++
 21 files changed, 250 insertions(+), 55 deletions(-)

diff --git a/bpkg/archive.cxx b/bpkg/archive.cxx
index 1471029..c096701 100644
--- a/bpkg/archive.cxx
+++ b/bpkg/archive.cxx
@@ -3,13 +3,10 @@
 
 #include <bpkg/archive.hxx>
 
-#include <libbutl/process.mxx>
-
 #include <bpkg/utility.hxx>
 #include <bpkg/diagnostics.hxx>
 
 using namespace std;
-using namespace butl;
 
 namespace bpkg
 {
diff --git a/bpkg/archive.hxx b/bpkg/archive.hxx
index f46812f..cfc4cdb 100644
--- a/bpkg/archive.hxx
+++ b/bpkg/archive.hxx
@@ -4,8 +4,6 @@
 #ifndef BPKG_ARCHIVE_HXX
 #define BPKG_ARCHIVE_HXX
 
-#include <libbutl/process.mxx>
-
 #include <bpkg/types.hxx>
 #include <bpkg/utility.hxx>
 
diff --git a/bpkg/auth.cxx b/bpkg/auth.cxx
index e29fdb9..e661ad0 100644
--- a/bpkg/auth.cxx
+++ b/bpkg/auth.cxx
@@ -9,7 +9,6 @@
 
 #include <libbutl/sha256.mxx>
 #include <libbutl/base64.mxx>
-#include <libbutl/process.mxx>
 #include <libbutl/openssl.mxx>
 #include <libbutl/timestamp.mxx>
 #include <libbutl/filesystem.mxx>
diff --git a/bpkg/checksum.cxx b/bpkg/checksum.cxx
index 228e0c5..efc86dd 100644
--- a/bpkg/checksum.cxx
+++ b/bpkg/checksum.cxx
@@ -7,13 +7,9 @@
 #  include <algorithm> // replace()
 #endif
 
-#include <libbutl/process.mxx>
-#include <libbutl/filesystem.mxx>
-
 #include <bpkg/diagnostics.hxx>
 
 using namespace std;
-using namespace butl;
 
 namespace bpkg
 {
diff --git a/bpkg/common.cli b/bpkg/common.cli
index c94f250..4537961 100644
--- a/bpkg/common.cli
+++ b/bpkg/common.cli
@@ -181,12 +181,30 @@ namespace bpkg
        programs."
     }
 
+    butl::url --pkg-proxy
+    {
+      "<url>",
+      "HTTP proxy server to use when fetching package manifests and archives
+       from remote \cb{pkg} repositories. If specified, the proxy <url> must
+       be in the \c{\b{http://}\i{host}[\b{:}\i{port}]} form. If \ci{port} is
+       omitted, 80 is used by default.
+
+       Note that to allow caching, the proxied \cb{https://} URLs are
+       converted to \cb{http://} in order to prevent the fetch program from
+       tunneling (which is the standard approach for proxying HTTPS). If both
+       HTTP and HTTPS repositories are used, it is assumed that the proxy
+       server can figure out which URLs need to be converted back to
+       \cb{https://} based on the request information (for example, host
+       name). For security, this mechanism should only be used with signed
+       repositories or when the proxy is located inside a trusted network."
+    }
+
     path --git = "git"
     {
       "<path>",
       "The git program to be used to fetch git repositories. You can also
-       specify additional options that should be passed to the git program with
-       \cb{--git-option}.
+       specify additional options that should be passed to the git program
+       with \cb{--git-option}.
 
        If the git program is not explicitly specified, then \cb{bpkg} will use
        \cb{git} by default."
diff --git a/bpkg/diagnostics.cxx b/bpkg/diagnostics.cxx
index 5b77e65..a471d07 100644
--- a/bpkg/diagnostics.cxx
+++ b/bpkg/diagnostics.cxx
@@ -7,8 +7,8 @@
 
 #include <odb/statement.hxx>
 
-#include <libbutl/process.mxx>
-#include <libbutl/process-io.mxx> // operator<<(ostream, process_arg)
+#include <libbutl/process.mxx>    // process_args
+#include <libbutl/process-io.mxx> // operator<<(ostream, process_*)
 
 #include <bpkg/utility.hxx>
 
@@ -22,14 +22,31 @@ namespace bpkg
   void
   print_process (const char* const args[], size_t n)
   {
-    diag_record r (text);
-    print_process (r, args, n);
+    diag_record dr (text);
+    print_process (dr, args, n);
   }
 
   void
-  print_process (diag_record& r, const char* const args[], size_t n)
+  print_process (diag_record& dr, const char* const args[], size_t n)
   {
-    r << process_args {args, n};
+    dr << process_args {args, n};
+  }
+
+  void
+  print_process (const process_env& pe, const char* const args[], size_t n)
+  {
+    diag_record dr (text);
+    print_process (dr, pe, args, n);
+  }
+
+  void
+  print_process (diag_record& dr,
+                 const process_env& pe, const char* const args[], size_t n)
+  {
+    if (pe.env ())
+      dr << pe << ' ';
+
+    dr << process_args {args, n};
   }
 
   // Diagnostics verbosity level.
diff --git a/bpkg/diagnostics.hxx b/bpkg/diagnostics.hxx
index ac19a1b..d11ab0b 100644
--- a/bpkg/diagnostics.hxx
+++ b/bpkg/diagnostics.hxx
@@ -49,6 +49,28 @@ namespace bpkg
     print_process (args.data (), args.size ());
   }
 
+  // As above but with process_env.
+  //
+  void
+  print_process (diag_record&,
+                 const process_env&, const char* const args[], size_t n = 0);
+
+  void
+  print_process (const process_env&, const char* const args[], size_t n = 0);
+
+  inline void
+  print_process (diag_record& dr,
+                 const process_env& pe, const cstrings& args)
+  {
+    print_process (dr, pe, args.data (), args.size ());
+  }
+
+  inline void
+  print_process (const process_env& pe, const cstrings& args)
+  {
+    print_process (pe, args.data (), args.size ());
+  }
+
   // Verbosity level. Update documentation for --verbose if changing.
   //
   // 0 - disabled
diff --git a/bpkg/fetch-git.cxx b/bpkg/fetch-git.cxx
index 274108f..df59dde 100644
--- a/bpkg/fetch-git.cxx
+++ b/bpkg/fetch-git.cxx
@@ -8,7 +8,6 @@
 
 #include <libbutl/git.mxx>
 #include <libbutl/utility.mxx>          // digit(), xdigit()
-#include <libbutl/process.mxx>
 #include <libbutl/filesystem.mxx>       // path_entry
 #include <libbutl/path-pattern.mxx>
 #include <libbutl/semantic-version.mxx>
diff --git a/bpkg/fetch-pkg.cxx b/bpkg/fetch-pkg.cxx
index 67ee3a9..81d4131 100644
--- a/bpkg/fetch-pkg.cxx
+++ b/bpkg/fetch-pkg.cxx
@@ -24,7 +24,11 @@ namespace bpkg
                   bool ignore_unknown)
   {
     string url (u.string ());
-    process pr (start_fetch (o, url));
+    process pr (start_fetch (o,
+                             url,
+                             path () /* out */,
+                             string () /* user_agent */,
+                             o.pkg_proxy ()));
 
     try
     {
@@ -94,7 +98,11 @@ namespace bpkg
       fail << "file " << df << " already exists";
 
     auto_rmfile arm (df);
-    process pr (start_fetch (o, u.string (), df));
+    process pr (start_fetch (o,
+                             u.string (),
+                             df,
+                             string () /* user_agent */,
+                             o.pkg_proxy ()));
 
     if (!pr.wait ())
     {
diff --git a/bpkg/fetch.cxx b/bpkg/fetch.cxx
index 63a459d..f565f26 100644
--- a/bpkg/fetch.cxx
+++ b/bpkg/fetch.cxx
@@ -3,12 +3,9 @@
 
 #include <bpkg/fetch.hxx>
 
-#include <libbutl/process.mxx>
-
 #include <bpkg/diagnostics.hxx>
 
 using namespace std;
-using namespace butl;
 
 namespace bpkg
 {
@@ -94,7 +91,8 @@ namespace bpkg
               const strings& ops,
               const string& url,
               const path& out,
-              const string& user_agent)
+              const string& user_agent,
+              const string& http_proxy)
   {
     bool fo (!out.empty ()); // Output to file.
 
@@ -178,10 +176,23 @@ namespace bpkg
     args.push_back (url.c_str ());
     args.push_back (nullptr);
 
-    process_path pp (process::path_search (args[0]));
+    process_path pp  (process::path_search (args[0]));
+    process_env  env (pp);
+
+    // HTTP proxy.
+    //
+    string evar;
+    const char* evars[] = {nullptr, nullptr};
+
+    if (!http_proxy.empty ())
+    {
+      evar = "http_proxy=" + http_proxy;
+      evars[0] = evar.c_str ();
+      env.vars = evars;
+    }
 
     if (verb >= 2)
-      print_process (args);
+      print_process (env, args);
 
     // If we are fetching into a file, change the wget's directory to
     // that of the output file. We do it this way so that we end up with
@@ -191,8 +202,9 @@ namespace bpkg
     return fo
       ? process (pp, args.data (),
                  0, 1, 2,
-                 out.directory ().string ().c_str ())
-      : process (pp, args.data (), 0, -1);
+                 out.directory ().string ().c_str (),
+                 env.vars)
+      : process (pp, args.data (), 0, -1, 2, nullptr /* cwd */, env.vars);
   }
 
   // curl
@@ -247,7 +259,8 @@ namespace bpkg
               const strings& ops,
               const string& url,
               const path& out,
-              const string& user_agent)
+              const string& user_agent,
+              const string& http_proxy)
   {
     bool fo (!out.empty ()); // Output to file.
 
@@ -320,6 +333,14 @@ namespace bpkg
       args.push_back (out.string ().c_str ());
     }
 
+    // HTTP proxy.
+    //
+    if (!http_proxy.empty ())
+    {
+      args.push_back ("--proxy");
+      args.push_back (http_proxy.c_str ());
+    }
+
     args.push_back (url.c_str ());
     args.push_back (nullptr);
 
@@ -396,7 +417,8 @@ namespace bpkg
                const strings& ops,
                const string& url,
                const path& out,
-               const string& user_agent)
+               const string& user_agent,
+               const string& http_proxy)
   {
     bool fo (!out.empty ()); // Output to file.
 
@@ -464,10 +486,23 @@ namespace bpkg
     args.push_back (url.c_str ());
     args.push_back (nullptr);
 
-    process_path pp (process::path_search (args[0]));
+    process_path pp  (process::path_search (args[0]));
+    process_env  env (pp);
+
+    // HTTP proxy.
+    //
+    string evar;
+    const char* evars[] = {nullptr, nullptr};
+
+    if (!http_proxy.empty ())
+    {
+      evar = "HTTP_PROXY=" + http_proxy;
+      evars[0] = evar.c_str ();
+      env.vars = evars;
+    }
 
     if (verb >= 2)
-      print_process (args);
+      print_process (env, args);
 
     // If we are fetching into a file, change the fetch's directory to
     // that of the output file. We do it this way so that we end up with
@@ -477,8 +512,9 @@ namespace bpkg
     return fo
       ? process (pp, args.data (),
                  0, 1, 2,
-                 out.directory ().string ().c_str ())
-      : process (pp, args.data (), 0, -1);
+                 out.directory ().string ().c_str (),
+                 env.vars)
+      : process (pp, args.data (), 0, -1, 2, nullptr /* cwd */, env.vars);
   }
 
   // The dispatcher.
@@ -572,9 +608,10 @@ namespace bpkg
 
   process
   start_fetch (const common_options& o,
-               const string& url,
+               const string& src,
                const path& out,
-               const string& user_agent)
+               const string& user_agent,
+               const url& proxy)
   {
     process (*f) (const path&,
                   const optional<size_t>&,
@@ -582,6 +619,7 @@ namespace bpkg
                   const strings&,
                   const string&,
                   const path&,
+                  const string&,
                   const string&) = nullptr;
 
     switch (check (o))
@@ -595,15 +633,79 @@ namespace bpkg
     if (o.fetch_timeout_specified ())
       timeout = o.fetch_timeout ();
 
+    // If the HTTP proxy is specified and the URL is HTTP(S), then fetch
+    // through the proxy, converting the https URL scheme to http.
+    //
     try
     {
+      string http_url;
+      string http_proxy;
+
+      if (!proxy.empty ())
+      {
+        auto bad_proxy = [&src, &proxy] (const char* d)
+        {
+          fail << "unable to fetch '" << src << "' using '" << proxy
+               << "' as proxy: " << d;
+        };
+
+        if (icasecmp (proxy.scheme, "http") != 0)
+          bad_proxy ("only HTTP proxy is supported");
+
+        if (!proxy.authority || proxy.authority->host.empty ())
+          bad_proxy ("invalid host name in proxy URL");
+
+        if (!proxy.authority->user.empty ())
+          bad_proxy ("unexpected user in proxy URL");
+
+        if (proxy.path)
+          bad_proxy ("unexpected path in proxy URL");
+
+        if (proxy.query)
+          bad_proxy ("unexpected query in proxy URL");
+
+        if (proxy.fragment)
+          bad_proxy ("unexpected fragment in proxy URL");
+
+        if (proxy.rootless)
+          bad_proxy ("proxy URL cannot be rootless");
+
+        url u;
+        try
+        {
+          u = url (src);
+        }
+        catch (const invalid_argument& e)
+        {
+          fail << "unable to fetch '" << src << "': invalid URL: " << e;
+        }
+
+        bool http  (icasecmp (u.scheme, "http")  == 0);
+        bool https (icasecmp (u.scheme, "https") == 0);
+
+        if (http || https)
+        {
+          http_proxy = proxy.string ();
+
+          if (proxy.authority->port == 0)
+            http_proxy += ":80";
+
+          if (https)
+          {
+            u.scheme = "http";
+            http_url = u.string ();
+          }
+        }
+      }
+
       return f (fetch_path,
                 timeout,
                 o.no_progress (),
                 o.fetch_option (),
-                url,
+                !http_url.empty () ? http_url : src,
                 out,
-                user_agent);
+                user_agent,
+                http_proxy);
     }
     catch (const process_error& e)
     {
diff --git a/bpkg/fetch.hxx b/bpkg/fetch.hxx
index ed1fd50..d57dcf3 100644
--- a/bpkg/fetch.hxx
+++ b/bpkg/fetch.hxx
@@ -6,8 +6,6 @@
 
 #include <ctime> // time_t
 
-#include <libbutl/process.mxx>
-
 #include <libbpkg/manifest.hxx>
 
 #include <bpkg/types.hxx>
@@ -20,6 +18,9 @@ namespace bpkg
   // Repository type pkg (fetch-pkg.cxx).
   //
 
+  // If HTTP proxy is specified via the --pkg-proxy option, then use it for
+  // fetching manifests and archives from the remote pkg repository.
+  //
   pkg_repository_manifests
   pkg_fetch_repositories (const dir_path&, bool ignore_unknown);
 
@@ -120,13 +121,17 @@ namespace bpkg
   // Start the process of fetching the specified URL. If out is empty, then
   // fetch to stdout. In this case also don't show any progress unless we are
   // running verbose. If user_agent is empty, then send the default (fetch
-  // program specific) User-Agent header value.
+  // program specific) User-Agent header value. If the HTTP proxy URL is not
+  // empty and the URL to fetch is HTTP(S), then fetch it via the specified
+  // proxy server converting the https URL scheme to http (see the --pkg-proxy
+  // option for details).
   //
-  butl::process
+  process
   start_fetch (const common_options& o,
                const string& url,
                const path& out = {},
-               const string& user_agent = {});
+               const string& user_agent = {},
+               const butl::url& proxy = {});
 }
 
 #endif // BPKG_FETCH_HXX
diff --git a/bpkg/manifest-utility.cxx b/bpkg/manifest-utility.cxx
index 697f920..159b2c3 100644
--- a/bpkg/manifest-utility.cxx
+++ b/bpkg/manifest-utility.cxx
@@ -6,7 +6,6 @@
 #include <cstring> // strcspn()
 
 #include <libbutl/b.mxx>
-#include <libbutl/url.mxx>
 #include <libbutl/sha256.mxx>
 
 #include <bpkg/package.hxx>        // wildcard_version
diff --git a/bpkg/pkg-build.cxx b/bpkg/pkg-build.cxx
index 6b68a65..106666a 100644
--- a/bpkg/pkg-build.cxx
+++ b/bpkg/pkg-build.cxx
@@ -10,7 +10,6 @@
 #include <iostream>   // cout
 #include <algorithm>  // find_if()
 
-#include <libbutl/url.mxx>
 #include <libbutl/standard-version.mxx>
 
 #include <bpkg/package.hxx>
@@ -2645,7 +2644,7 @@ namespace bpkg
         //
         size_t p (0);
 
-        using url_traits = butl::url::traits_type;
+        using url_traits = url::traits_type;
 
         // Skip leading ':' that are not part of a URL.
         //
diff --git a/bpkg/pkg-unpack.cxx b/bpkg/pkg-unpack.cxx
index 02d61b6..10f8920 100644
--- a/bpkg/pkg-unpack.cxx
+++ b/bpkg/pkg-unpack.cxx
@@ -7,8 +7,6 @@
 #  include <algorithm> // replace()
 #endif
 
-#include <libbutl/process.mxx>
-
 #include <libbpkg/manifest.hxx>
 
 #include <bpkg/archive.hxx>
@@ -23,7 +21,6 @@
 #include <bpkg/pkg-verify.hxx>
 
 using namespace std;
-using namespace butl;
 
 namespace bpkg
 {
diff --git a/bpkg/pkg-verify.cxx b/bpkg/pkg-verify.cxx
index 5c23b69..38b4d68 100644
--- a/bpkg/pkg-verify.cxx
+++ b/bpkg/pkg-verify.cxx
@@ -5,7 +5,6 @@
 
 #include <iostream> // cout
 
-#include <libbutl/process.mxx>
 #include <libbutl/manifest-parser.mxx>
 #include <libbutl/manifest-serializer.mxx>
 
diff --git a/bpkg/satisfaction.cxx b/bpkg/satisfaction.cxx
index c2a4601..52def32 100644
--- a/bpkg/satisfaction.cxx
+++ b/bpkg/satisfaction.cxx
@@ -3,13 +3,10 @@
 
 #include <bpkg/satisfaction.hxx>
 
-#include <libbutl/process.mxx>
-
 #include <bpkg/package-odb.hxx>
 #include <bpkg/diagnostics.hxx>
 
 using namespace std;
-using namespace butl;
 
 namespace bpkg
 {
diff --git a/bpkg/types-parsers.cxx b/bpkg/types-parsers.cxx
index ce3cd0c..be95219 100644
--- a/bpkg/types-parsers.cxx
+++ b/bpkg/types-parsers.cxx
@@ -7,6 +7,28 @@ namespace bpkg
 {
   namespace cli
   {
+    void parser<url>::
+    parse (url& x, bool& xs, scanner& s)
+    {
+      const char* o (s.next ());
+
+      if (!s.more ())
+        throw missing_value (o);
+
+      const char* v (s.next ());
+
+      try
+      {
+        x = url (v);
+      }
+      catch (const invalid_argument& e)
+      {
+        throw invalid_value (o, v, e.what ());
+      }
+
+      xs = true;
+    }
+
     template <typename T>
     static void
     parse_path (T& x, scanner& s)
diff --git a/bpkg/types-parsers.hxx b/bpkg/types-parsers.hxx
index c812e7b..38b7cee 100644
--- a/bpkg/types-parsers.hxx
+++ b/bpkg/types-parsers.hxx
@@ -19,6 +19,16 @@ namespace bpkg
   namespace cli
   {
     template <>
+    struct parser<url>
+    {
+      static void
+      parse (url&, bool&, scanner&);
+
+      static void
+      merge (url& b, const url& a) {b = a;}
+    };
+
+    template <>
     struct parser<path>
     {
       static void
diff --git a/bpkg/types.hxx b/bpkg/types.hxx
index d4847b8..4f0cc98 100644
--- a/bpkg/types.hxx
+++ b/bpkg/types.hxx
@@ -21,6 +21,7 @@
 
 #include <odb/lazy-ptr.hxx>
 
+#include <libbutl/url.mxx>
 #include <libbutl/path.mxx>
 #include <libbutl/process.mxx>
 #include <libbutl/utility.mxx>         // compare_reference_target
@@ -94,9 +95,14 @@ namespace bpkg
   using paths = std::vector<path>;
   using dir_paths = std::vector<dir_path>;
 
+  // <libbutl/url.mxx>
+  //
+  using butl::url;
+
   // <libbutl/process.mxx>
   //
   using butl::process;
+  using butl::process_env;
   using butl::process_path;
   using butl::process_exit;
   using butl::process_error;
@@ -104,6 +110,7 @@ namespace bpkg
   // <libbutl/fdstream.mxx>
   //
   using butl::auto_fd;
+  using butl::nullfd;
   using butl::fdpipe;
   using butl::ifdstream;
   using butl::ofdstream;
diff --git a/bpkg/utility.cxx b/bpkg/utility.cxx
index 5ec8159..9c9c991 100644
--- a/bpkg/utility.cxx
+++ b/bpkg/utility.cxx
@@ -4,7 +4,6 @@
 #include <bpkg/utility.hxx>
 
 #include <libbutl/prompt.mxx>
-#include <libbutl/process.mxx>
 #include <libbutl/fdstream.mxx>
 
 #include <bpkg/diagnostics.hxx>
diff --git a/bpkg/utility.hxx b/bpkg/utility.hxx
index 1763af0..e891f21 100644
--- a/bpkg/utility.hxx
+++ b/bpkg/utility.hxx
@@ -14,6 +14,7 @@
 #include <libbutl/ft/lang.hxx>
 
 #include <libbutl/utility.mxx>         // icasecmp(), reverse_iterate(), etc
+#include <libbutl/process.mxx>
 #include <libbutl/filesystem.mxx>
 #include <libbutl/default-options.mxx>
 
@@ -47,6 +48,10 @@ namespace bpkg
   using butl::setenv;
   using butl::unsetenv;
 
+  // <libbutl/process.mxx>
+  //
+  using butl::process_start_callback;
+
   // <libbutl/filesystem.mxx>
   //
   using butl::auto_rmfile;
-- 
cgit v1.1