From 0f29c103d9992ca2ae9336c1fa18da98234e800c Mon Sep 17 00:00:00 2001
From: Chris Bradley <chris.bradley@ecmwf.int>
Date: Tue, 26 Nov 2024 17:01:27 +0000
Subject: [PATCH] Improve scan command line arg support

---
 src/tools/gribjump-scan.cc         | 66 +++++++++++++++++++-----------
 tests/tools/callback_vs_scan.sh.in |  2 +-
 2 files changed, 44 insertions(+), 24 deletions(-)

diff --git a/src/tools/gribjump-scan.cc b/src/tools/gribjump-scan.cc
index 0d601fe..ceb370d 100644
--- a/src/tools/gribjump-scan.cc
+++ b/src/tools/gribjump-scan.cc
@@ -15,6 +15,7 @@
 #include "metkit/mars/MarsExpension.h"
 
 #include "fdb5/api/FDB.h"
+#include "fdb5/api/helpers/FDBToolRequest.h"
 
 #include "gribjump/GribJump.h"
 #include "gribjump/tools/GribJumpTool.h"
@@ -31,21 +32,21 @@ class Scan: public GribJumpTool {
     
     virtual void execute(const eckit::option::CmdArgs &args);
     virtual void usage(const std::string &tool) const;
-    virtual int numberOfPositionalArguments() const { return 1; }
+    virtual int numberOfPositionalArguments() const { return -1; }
   
   public:
     Scan(int argc, char **argv): GribJumpTool(argc, argv) {
+        options_.push_back(new eckit::option::SimpleOption<std::string>("file", "Reads the mars requests from a file, rather than from the command line"));
         options_.push_back(new eckit::option::SimpleOption<bool>("raw", "Uses the raw request, without expansion"));
-        options_.push_back(new eckit::option::SimpleOption<bool>("files", "Scan entire files matching the request (default: true)"));
+        options_.push_back(new eckit::option::SimpleOption<bool>("byfiles", "Scan entire files matching the request (default: true)"));
     }
 
 };
 
 void Scan::usage(const std::string &tool) const {
     eckit::Log::info() << std::endl
-                       << "Usage: " << tool << " <mars request file>" << std::endl
-                       << "       " << tool << " --raw <mars request file>" << std::endl
-                       << "       " << tool << " --files <mars request file>" << std::endl
+                       << "Usage: " << tool << " class=od,stream=oper,expver=xxxx" << std::endl
+                       << "       " << tool << " --file=<mars request file>" << std::endl
                        ;
 
     GribJumpTool::usage(tool);
@@ -54,29 +55,48 @@ void Scan::usage(const std::string &tool) const {
 void Scan::execute(const eckit::option::CmdArgs &args) {
 
     bool raw = args.getBool("raw", false);
-    bool files = args.getBool("files", false);
-
-    // Build request(s) from input
-    std::ifstream in(args(0).c_str());
-    if (in.bad()) {
-        throw eckit::ReadError(args(0));
-    }
+    bool byfiles = args.getBool("byfiles", false);
+    std::string file = args.getString("file", "");
 
     std::vector<metkit::mars::MarsRequest> requests;
-    metkit::mars::MarsParser parser(in);
-    auto parsedRequests = parser.parse();
-    if (raw) {
-        for (auto r : parsedRequests)
-            requests.push_back(r);
+    
+    if (args.count() > 0 && !file.empty()) {
+        throw eckit::UserError("Invalid arguments: Cannot specify both a file (--file) and a request (positional arguments)");
+    }
+    
+    if (!file.empty()){
+        // Build request(s) from <file>
+        std::ifstream in(file);
+        if (in.bad()) {
+            throw eckit::ReadError(args(0));
+        }
+
+        metkit::mars::MarsParser parser(in);
+        auto parsedRequests = parser.parse();
+        if (raw) {
+            for (auto r : parsedRequests)
+                requests.push_back(r);
+        } else {
+            bool inherit = false;
+            metkit::mars::MarsExpension expand(inherit);
+            requests = expand.expand(parsedRequests);
+        }
     } else {
-        bool inherit = false;
-        metkit::mars::MarsExpension expand(inherit);
-        requests = expand.expand(parsedRequests);
+        // Build request(s) from positional arguments
+        std::vector<std::string> requests_in;
+        for (size_t i = 0; i < args.count(); ++i) {
+            requests_in.emplace_back(args(i));
+        }
+
+        for (const std::string& request_string : requests_in) {
+            auto parsed = fdb5::FDBToolRequest::requestsFromString(request_string, {}, raw, "retrieve");
+            for (auto r : parsed) {
+                requests.push_back(r.request());
+            }
+        }
     }
-
     GribJump gj;
-
-    size_t nfields = gj.scan(requests, files);
+    size_t nfields = gj.scan(requests, byfiles);
     eckit::Log::info() << "Scanned " << nfields << " field(s)" << std::endl;
 }
 
diff --git a/tests/tools/callback_vs_scan.sh.in b/tests/tools/callback_vs_scan.sh.in
index 51db402..46ade95 100755
--- a/tests/tools/callback_vs_scan.sh.in
+++ b/tests/tools/callback_vs_scan.sh.in
@@ -80,7 +80,7 @@ FDB5_CONFIG_FILE=${bindir}/config_root1.yaml $fdbwrite ${testdata}
 
 echo "Writing to FDB without gribjump plugin, then scanning"
 FDB5_CONFIG_FILE=${bindir}/config_root2.yaml $fdbwrite ${testdata}
-FDB5_CONFIG_FILE=${bindir}/config_root2.yaml $gjscan $selectedrequests
+FDB5_CONFIG_FILE=${bindir}/config_root2.yaml $gjscan --file=$selectedrequests
 
 # -------------------------------------
 # 3: FDB with gribjump plugin