rebrand: metha

.gitignore

@@ -1 +1 @@
-/perimorph-*
+/metha-*

Makefile

@@ -1,7 +1,7 @@
 SHELL = /bin/bash
-TARGETS = perimorph-cat perimorph-info perimorph-records perimorph-sync
+TARGETS = metha-sync
 
-PKGNAME = perimorph
+PKGNAME = metha
 
 all: $(TARGETS)
 

README.md

@@ -1,24 +1,22 @@
-Perimorph
-=========
+metha
+=====
 
-> A mineral enclosed within another.
-
-Command line OAI-PMH client and incremental harvester.
+Command line OAI-PMH incremental harvester.
 
 Usage
 -----
 
 ```sh
-$ perimorph-info http://export.arxiv.org/oai2
+$ metha-info http://export.arxiv.org/oai2
 {
-	"identify": ...,
-	"formats": ...,
-	"sets": ...
+    "identify": ...,
+    "formats": ...,
+    "sets": ...
 }
-$ perimorph-sync http://export.arxiv.org/oai2
+$ metha-sync http://export.arxiv.org/oai2
 ...
-$ perimorph-cat http://export.arxiv.org/oai2
+$ metha-cat http://export.arxiv.org/oai2
 ...
-$ perimorph-records http://export.arxiv.org/oai2
+$ metha-records http://export.arxiv.org/oai2
 ...
 ```

client.go

@@ -1,4 +1,4 @@
-package perimorph
+package metha
 
 import (
 	"bytes"
@@ -20,17 +20,9 @@ const (
 	DefaultMaxRetries = 8
 )
 
-// defaultDoer is a resilient http client with retries and longer timeout.
-var defaultDoer = func() Doer {
-	c := pester.New()
-	c.Timeout = DefaultTimeout
-	c.MaxRetries = DefaultMaxRetries
-	c.Backoff = pester.ExponentialBackoff
-	return c
-}()
-
 var (
-	DefaultClient = Client{Doer: defaultDoer}
+	StdClient     = Client{Doer: http.DefaultClient}
+	DefaultClient = Client{Doer: CreateDoer(DefaultTimeout, DefaultMaxRetries)}
 	// Example for broken XML: http://eprints.vu.edu.au/perl/oai2. Add more
 	// weird things to be cleaned before XML parsing here. Another faulty:
 	// http://digitalcommons.gardner-webb.edu/do/oai/?from=2016-02-29&metadataPr
@@ -49,6 +41,24 @@ var (
 		"\u001E", "", "\u001F", "")
 )
 
+// CreateDoer will return http request clients with specific timeout and retry
+// properties.
+func CreateDoer(timeout time.Duration, retries int) Doer {
+	if timeout == 0 && retries == 0 {
+		return http.DefaultClient
+	}
+	c := pester.New()
+	c.Timeout = timeout
+	c.MaxRetries = retries
+	c.Backoff = pester.ExponentialBackoff
+	return c
+}
+
+// Create a client with timeout and retry properties.
+func CreateClient(timeout time.Duration, retries int) Client {
+	return Client{Doer: CreateDoer(timeout, retries)}
+}
+
 // Doer is a minimal HTTP interface.
 type Doer interface {
 	Do(*http.Request) (*http.Response, error)

cmd/perimorph-sync/main.go → cmd/metha-sync/main.go

@@ -6,7 +6,7 @@ import (
 	"log"
 	"os"
 
-	"github.com/miku/perimorph"
+	"github.com/miku/metha"
 )
 
 func main() {
@@ -21,15 +21,29 @@ func main() {
 	flag.Parse()
 
 	if *version {
-		fmt.Println(perimorph.Version)
+		fmt.Println(metha.Version)
 		os.Exit(0)
 	}
 
 	if flag.NArg() == 0 {
 		log.Fatal("endpoint required")
 	}
 
-	harvest, err := perimorph.NewHarvest(perimorph.PrependSchema(flag.Arg(0)))
+	baseURL := metha.PrependSchema(flag.Arg(0))
+
+	if *showDir {
+		// showDir only needs these parameters
+		harvest := metha.Harvest{
+			BaseURL: baseURL,
+			Format:  *format,
+			Set:     *set,
+		}
+		fmt.Println(harvest.Dir())
+		os.Exit(0)
+	}
+
+	// NewHarvest ensures the endpoint is sane, before we start
+	harvest, err := metha.NewHarvest(baseURL)
 	if err != nil {
 		log.Fatal(err)
 	}
@@ -41,11 +55,6 @@ func main() {
 	harvest.DisableSelectiveHarvesting = *disableSelectiveHarvesting
 	harvest.MaxEmptyResponses = 10
 
-	if *showDir {
-		fmt.Println(harvest.Dir())
-		os.Exit(0)
-	}
-
 	log.Printf("harvest: %+v", harvest)
 
 	if err := harvest.Run(); err != nil {

fileutil.go

@@ -1,4 +1,4 @@
-package perimorph
+package metha
 
 import "path/filepath"
 

harvest.go

@@ -1,4 +1,4 @@
-package perimorph
+package metha
 
 import (
 	"encoding/base64"
@@ -22,7 +22,7 @@ const Day = 24 * time.Hour
 
 var (
 	// BaseDir is where all downloaded data is stored
-	BaseDir     = filepath.Join(UserHomeDir(), ".perimorph")
+	BaseDir     = filepath.Join(UserHomeDir(), ".metha")
 	fnPattern   = regexp.MustCompile("(?P<Date>[0-9]{4,4}-[0-9]{2,2}-[0-9]{2,2})-[0-9]{8,}.xml(.gz)?$")
 	datePattern = regexp.MustCompile("[0-9]{4,4}-[0-9]{2,2}-[0-9]{2,2}")
 
@@ -354,7 +354,11 @@ func (h *Harvest) earliestDate() (time.Time, error) {
 
 func (h *Harvest) identify() error {
 	req := Request{Verb: "Identify", BaseURL: h.BaseURL}
-	resp, err := Do(&req)
+
+	// use a less resilient client for indentify requests
+	c := CreateClient(1*time.Second, 2)
+
+	resp, err := c.Do(&req)
 	if err != nil {
 		return err
 	}
@@ -363,7 +367,7 @@ func (h *Harvest) identify() error {
 }
 
 func init() {
-	if dir := os.Getenv("PERIMORPH_DIR"); dir != "" {
+	if dir := os.Getenv("METHA_DIR"); dir != "" {
 		BaseDir = dir
 	}
 }

home.go

@@ -1,4 +1,4 @@
-package perimorph
+package metha
 
 import (
 	"os"

intervals.go

@@ -1,4 +1,4 @@
-package perimorph
+package metha
 
 import (
 	"time"

laster.go

@@ -1,4 +1,4 @@
-package perimorph
+package metha
 
 import (
 	"io/ioutil"

logging.go

@@ -1,4 +1,4 @@
-package perimorph
+package metha
 
 import (
 	"flag"

multierror.go

@@ -1,4 +1,4 @@
-package perimorph
+package metha
 
 import (
 	"bytes"

request.go

@@ -1,4 +1,4 @@
-package perimorph
+package metha
 
 import (
 	"bytes"

request_test.go

@@ -1,4 +1,4 @@
-package perimorph
+package metha
 
 import (
 	"net/url"

response.go

@@ -1,4 +1,4 @@
-package perimorph
+package metha
 
 import (
 	"bytes"

version.go

@@ -1,3 +1,3 @@
-package perimorph
+package metha
 
 const Version = "0.1.0"