From 3e787f8d19037fbcad0a9486dc1eb2a1f6dc1daf Mon Sep 17 00:00:00 2001
From: Ian Rees <ian@ianrees.net>
Date: Tue, 12 May 2020 20:00:03 -0700
Subject: [PATCH] Expose additional copier options to dmfr import (#55)

---
 copier/copier.go   |  2 +-
 dmfr/import.go     | 20 +++++++++++--------
 dmfr/import_cmd.go | 49 ++++++++++++++++++++++++----------------------
 3 files changed, 39 insertions(+), 32 deletions(-)

diff --git a/copier/copier.go b/copier/copier.go
index c7581f69..f65b355b 100644
--- a/copier/copier.go
+++ b/copier/copier.go
@@ -104,7 +104,7 @@ func NewCopier(reader gotransit.Reader, writer gotransit.Writer) Copier {
 	copier := Copier{
 		Reader:               reader,
 		Writer:               writer,
-		BatchSize:            100000,
+		BatchSize:            1000000,
 		AllowEntityErrors:    false,
 		AllowReferenceErrors: false,
 		InterpolateStopTimes: false,
diff --git a/dmfr/import.go b/dmfr/import.go
index a633b899..731caf77 100644
--- a/dmfr/import.go
+++ b/dmfr/import.go
@@ -16,11 +16,13 @@ import (
 
 // ImportOptions sets various options for importing a feed.
 type ImportOptions struct {
-	FeedVersionID int
-	Extensions    []string
-	Directory     string
-	S3            string
-	Activate      bool
+	FeedVersionID        int
+	Extensions           []string
+	Directory            string
+	S3                   string
+	Activate             bool
+	CreateMissingShapes  bool
+	InterpolateStopTimes bool
 }
 
 // ImportResult contains the results of a feed import.
@@ -220,12 +222,14 @@ func ImportFeedVersion(atx gtdb.Adapter, fv gotransit.FeedVersion, opts ImportOp
 		}
 		cp.AddExtension(ext)
 	}
-	cp.BatchSize = 1000000
+	// Settable options
+	cp.CreateMissingShapes = opts.CreateMissingShapes
+	cp.InterpolateStopTimes = opts.InterpolateStopTimes
+	// Non-settable options
 	cp.AllowEntityErrors = false
 	cp.AllowReferenceErrors = false
 	cp.NormalizeServiceIDs = true
-	cp.CreateMissingShapes = true
-	cp.InterpolateStopTimes = true
+	// Go
 	cpresult := cp.Copy()
 	if cpresult == nil {
 		return fvi, errors.New("copy result was nil")
diff --git a/dmfr/import_cmd.go b/dmfr/import_cmd.go
index 1f2f307a..bbe40e4c 100644
--- a/dmfr/import_cmd.go
+++ b/dmfr/import_cmd.go
@@ -13,46 +13,47 @@ import (
 
 // ImportCommand imports FeedVersions into a database.
 type ImportCommand struct {
-	S3           string
-	Directory    string
-	Activate     bool
-	Extensions   arrayFlags
-	Workers      int
-	Limit        int
-	DBURL        string
-	CoverDate    string
-	FetchedSince string
-	Latest       bool
-	DryRun       bool
-	FeedIDs      []string
-	FVIDs        arrayFlags
-	Adapter      gtdb.Adapter // allow for mocks
+	Workers       int
+	Limit         int
+	DBURL         string
+	CoverDate     string
+	FetchedSince  string
+	Latest        bool
+	DryRun        bool
+	FeedIDs       []string
+	FVIDs         arrayFlags
+	Adapter       gtdb.Adapter // allow for mocks
+	ImportOptions ImportOptions
 }
 
 // Parse command line flags
 func (cmd *ImportCommand) Parse(args []string) error {
+	extflags := arrayFlags{}
 	fl := flag.NewFlagSet("import", flag.ExitOnError)
 	fl.Usage = func() {
 		fmt.Println("Usage: import [feedids...]")
 		fl.PrintDefaults()
 	}
-	fl.Var(&cmd.Extensions, "ext", "Include GTFS Extension")
+	fl.Var(&extflags, "ext", "Include GTFS Extension")
 	fl.Var(&cmd.FVIDs, "fvid", "Import specific feed version ID")
 	fl.IntVar(&cmd.Workers, "workers", 1, "Worker threads")
 	fl.StringVar(&cmd.DBURL, "dburl", "", "Database URL (default: $DMFR_DATABASE_URL)")
-	fl.StringVar(&cmd.Directory, "gtfsdir", ".", "GTFS Directory")
-	fl.StringVar(&cmd.S3, "s3", "", "Get GTFS files from S3 bucket/prefix")
+	fl.StringVar(&cmd.ImportOptions.Directory, "gtfsdir", ".", "GTFS Directory")
+	fl.StringVar(&cmd.ImportOptions.S3, "s3", "", "Get GTFS files from S3 bucket/prefix")
 	fl.StringVar(&cmd.CoverDate, "date", "", "Service on date")
 	fl.StringVar(&cmd.FetchedSince, "fetched-since", "", "Fetched since")
 	fl.IntVar(&cmd.Limit, "limit", 0, "Import at most n feeds")
 	fl.BoolVar(&cmd.Latest, "latest", false, "Only import latest feed version available for each feed")
 	fl.BoolVar(&cmd.DryRun, "dryrun", false, "Dry run; print feeds that would be imported and exit")
-	fl.BoolVar(&cmd.Activate, "activate", false, "Set as active feed version after import")
+	fl.BoolVar(&cmd.ImportOptions.Activate, "activate", false, "Set as active feed version after import")
+	fl.BoolVar(&cmd.ImportOptions.InterpolateStopTimes, "interpolate-stop-times", false, "Interpolate missing StopTime arrival/departure values")
+	fl.BoolVar(&cmd.ImportOptions.CreateMissingShapes, "create-missing-shapes", false, "Create missing Shapes from Trip stop-to-stop geometries")
 	fl.Parse(args)
 	cmd.FeedIDs = fl.Args()
 	if cmd.DBURL == "" {
 		cmd.DBURL = os.Getenv("DMFR_DATABASE_URL")
 	}
+	cmd.ImportOptions.Extensions = extflags
 	return nil
 }
 
@@ -115,11 +116,13 @@ func (cmd *ImportCommand) Run() error {
 	results := make(chan ImportResult, len(qrs))
 	for _, fvid := range qrs {
 		jobs <- ImportOptions{
-			FeedVersionID: fvid,
-			Directory:     cmd.Directory,
-			S3:            cmd.S3,
-			Extensions:    cmd.Extensions,
-			Activate:      cmd.Activate,
+			FeedVersionID:        fvid,
+			Directory:            cmd.ImportOptions.Directory,
+			S3:                   cmd.ImportOptions.S3,
+			Extensions:           cmd.ImportOptions.Extensions,
+			Activate:             cmd.ImportOptions.Activate,
+			InterpolateStopTimes: cmd.ImportOptions.InterpolateStopTimes,
+			CreateMissingShapes:  cmd.ImportOptions.CreateMissingShapes,
 		}
 	}
 	close(jobs)