add the data prep by Greg

data/build-database

@@ -0,0 +1,3 @@
+#!/bin/bash
+rm -f bird_database.duckdb
+duckdb bird_database.duckdb < schema-build-script.sql

data/personnel.csv

@@ -0,0 +1,270 @@
+Abbreviation,Name
+agottesman,Aaron Gottesman
+asterling,Abby Sterling
+abradley,Adam Bradley
+akneidel,Alan Kneidel
+akataluk-primeau,Alannah Kataluk-Primeau
+aissisgaitok,Albert Issisgaitok
+amatsyna,Alexader Matsyna
+asoto,Alfredo Soto
+abuckley,Alvan Buckley
+ataqtaq,Amanda Taqtaq
+aguerra,Ana Guerra
+amines,Anaka Mines
+apopovkina,Anastasia Popovkina
+abankert,Andrew Bankert
+adoll,Andrew Doll
+aperry,Andrew Perry
+arbankert,Andrew R. Bankert
+asjohnson,Andrew S. Johnson
+abaltensperger,Andy Baltensperger
+ajohnson,Andy Johnson
+asmith,Angus Smith
+anja,Anja
+acorkery,Anne Corkery
+amould,Anthony Mould
+ancook,Ashley Nicole Cook
+atufts,Ashley Tufts
+amylnikova,Asya Mylnikova
+blalibert,Benoît Laliberté
+bkaselow,Billy Kaselow
+bwinn,Brad Winn
+bmwalker,Bradley M. Walker
+bwilkinson,Bradley Wilkinson
+bverheijen,Bram Verheijen
+bhiggins,Brendan Higgins
+bksandercock,Brett K. Sandercock
+bharrington,Brian Harrington
+brobinson,Brian Robinson
+bdalziel,Bronwyn Dalziel
+bhill,Brooke Hill
+cdavis,Caitlin Davis
+cbishop,Caitlyn Bishop
+cmorcos,Camila Morcos
+clahoda,Candace Lahoda
+cjrguerra,Carlos Jose Ruiz Guerra
+clishman,Carmen Lishman
+codwyer,Carol O'Dwyer
+cdoucet,Catherine Doucet
+cpham,Catherine Pham
+cwoodworth,Chelsea Woodworth
+cgratto-trevor,Cheri Gratto-Trevor
+chojnowski,Cheryl Hojnowski
+csmith,Chris Smith
+cbrunner,Christina Brunner
+canderson,Christine Anderson
+chelms,Clinton Helms
+cmurchison,Colleen Murchison
+dgough,Danielle Gough
+dedwards,Darryl Edwards
+dmcgeachy,Dave Mcgeachy
+dsaalfeld,Dave Saalfeld
+dattagutaluk,David Attagutaluk
+dblank,David B. Lank
+dhodkinson,David Hodkinson
+dpavlik,David Pavlik
+dward,David Ward
+dturner,Devin Turner
+dsolovyeva,Diana Solovyeva
+dborden,Diane Borden
+dfillman,Don Fillman
+djleandry,Don-Jean Léandry
+dmcrae,Doug McRae
+dkessler,Dylan Kessler
+ecorp,Edward Corp
+epierce,Elin Pierce
+epayne,Elizabeth Payne
+epero,Ellen Pero
+egirard,Eloise Girard
+echalifour,Émilie Chalifour
+edastrous,Emilie D'Astrous
+eweiser,Emily Weiser
+edavis,Emma Davis
+ereed,Eric Reed
+estroud,Eric Stroud
+eescajeda,Erica Escajeda
+enol,Erica Nol
+ekwon,Eunbi Kwon
+fylin,Fang-Yee Lin
+fsgagnon,Fanny Senez Gagnon
+fsanders,Felicia Sanders
+fsmith,Fletcher Smith
+frousseaux,Francois Rousseaux
+fstetler,Frank Stetler
+gmccune,Gennyne Mccune
+gpavlyukov,Georgiy Pavlyukov
+gracey,Gerry Racey
+gholmes,Gillian Holmes
+gdanilov,Gleb Danilov
+gsedash,Gleb Sedash
+ggilchrist,Grant Gilchrist
+gsolovyev,Grigory Solovyev
+gszor,Guillaume Szor
+hspecht,Hannah Specht
+hcraig,Heather Craig
+hhelling,Heidi Helling
+hbatcheller,Hope Batcheller
+idavies,Ian Davies
+ijong,Ian Jong
+jeschenroeder,Jackman Eschenroeder
+jkorol,Jaimlyn Korol
+jcaville,James Caville
+jhughey,Jared Hughey
+jmarin,Javier Marin
+jiron,Jean Iron
+jflamarre,Jean-François Lamarre
+jrjulian,Jean-Remi Julian
+jzamuido,Jeisson Zamuido
+jcoughlan,Jennifer Coughlan
+jgale,Jennifer Gale
+jkardiak,Jennifer Kardiak
+jcunningham,Jenny Cunningham
+jhixon,Jerrid Hixon
+jhupp,Jerry Hupp
+jrobinson,Jill Robinson
+jsauer,Jim Sauer
+jpanipak,Joanna Panipak
+jliebezeit,Joe Liebezeit
+jbety,Joël Bêty
+jperz,Johanna Perz
+jbrunjes,John Brunjes
+jdiener,John Diener
+jlautenbach,Jonathan Lautenbach
+jotis,José Otis
+jbelliveau,Julie Belliveau
+jdanby,Julie Danby
+jkelso,Julie Kelso
+jwebber,Julie Webber
+jloshchagina,Julya Loshchagina
+kaward,Kara-Anne Ward
+kreischke,Karissa Reischke
+kmatsyna,Katya Matsyna
+knuyaviak,Kayla Nuyaviak
+ksheimreif,Kayla Sheimreif
+kabraham,Ken Abraham
+kkalasz,Kevin Kalasz
+kkardynal,Kevin Kardynal
+kpietrzak,Kevin Pietrzak
+kbennett,Kim Bennett
+kjones,Kim Jones
+kgrond,Kirsten Grond
+kbeckmann,Kristina Beckmann
+kbertrand,Kyle Bertrand
+khogrefe,Kyle Hogrefe
+lkoloski,Laura Koloski
+lmckinnon,Laura McKinnon
+lpaquette,Laurence Paquette
+lrenzel,Leah Renzel
+lmercier,Léonie Mercier
+lyen,Leslie Yen
+lworing,Lewis W. Oring
+lbachellier,Lisa Bachellier
+lkennedy,Lisa Kennedy
+lpirie,Lisa Pirie
+lpollock,Lisa Pollock
+lgoodrick,Lizzie Goodrick
+ljolicoeur,Ludovic Jolicoeur
+lpadegimas,Lukas Padegimas
+lburlingame,Luke Burlingame
+mvheyden,Madeline Vander Heyden
+mmcconnell,Madison McConnell
+myyang,Man-Yu Yang (Mona)
+magiroux,Marie-Andrée Giroux
+mcfrenette,Marie-Christine Frenette
+mtrudel,Marion Trudel
+mdodds,Mark Dodds
+mgibson,Mark Gibson
+mlafaver,Mark Lafaver
+mpeck,Mark Peck
+mpatenaude-monette,Martin Patenaude-Monette
+mheung,Mary Heung
+mjones,Mary Jones
+mbirarda,Matt Birarda
+mmichaud,Matt Michaud
+msmith,Matt Smith
+mmudge,Mckenzie Mudge
+mmccloskey,Meagan Mccloskey
+mboldenow,Megan Boldenow
+mchown,Melanie Chown
+mvezina,Melanie Vezina
+mwilson,Melanie Wilson
+mmcgarvey,Metta Mcgarvey
+mbwunder,Michael B. Wunder
+mevers,Michael Evers
+mharing,Michaela Haring
+mballvanzee,Michelangelo Ballvanzee
+mallen,Mike Allen
+mburrell,Mike Burrell
+mqrunnut,Mike Qrunnut
+mettuvgii,Mikhail Ettuvgii
+mtrottier-paquet,Myriam Trottier-Paquet
+nmveld,Naomi Manin'T Veld
+nhentze,Nathan Hentze
+nkirby,Nathan Kirby
+nrsenner,Nathan R. Senner
+nlecomte,Nicolas Lecomte
+ncook,Nicole Cook
+nmacdonald,Nicole MacDonald
+nboulanger-lapointe,Noémie Boulanger-Lapointe
+ojames,Oliver James
+ohicks,Olivia Hicks
+oguy,Other Guy
+proyer-boutin,Pascal Royer-Boutin
+pherzog,Patrick Herzog
+psmith,Paul Smith
+pwoodard,Paul Woodard
+pchilton,Penelope Chilton
+pbertrand,Philippe Bertrand
+rhildebrandt,Reid Hildebrandt
+rlanctot,Richard Lanctot
+rgates,River Gates
+rnakoochee,Roberta Nakoochee
+rhunnewell,Robin Hunnewell
+radugan,Ronan A. Dugan
+rharned,Ronnie Harned
+rchurchwell,Roy Churchwell
+rhammond,Ruby Hammond
+rburner,Ryan Burner
+sfranks,Samantha Franks
+sjeswiet,Sarah Jeswiet
+sneima,Sarah Neima
+ssaalfeld,Sarah Saalfeld
+sflemming,Scott Flemming
+sfreeman,Scott Freeman
+svartanyan,Sergey Vartanyan
+scarvey,Shannon Carvey
+smbillerman,Shawn M. Billerman
+sskinner,Shelby Skinner
+sschulte,Shiloh Schulte
+sbuckell,Simon Buckell
+ssapora,Slade Sapora
+sbrown,Stephen Brown
+sparmiter,Stephen Parmiter
+sbennett,Steve Bennett
+shart,Steve Hart
+skendall,Steve Kendall
+skolbe,Steve Kolbe
+svissault,Steve Vissault
+ssheedy,Sydney Sheedy
+tdoyle,Terry Doyle
+tfarrugia,Thomas Farrugia
+triecke,Thomas Riecke
+tmankis,Tobias Mankis
+tstclair,Toby St Clair
+tesson,Tommy Esson
+tkydd,Tyler Kydd
+tdonnelly,Tyrone Donnelly
+vamarualik,Valerie Amarualik
+vloverti,Vanessa Loverti
+vbaranyuk,Vasily Baranyuk
+vkokhanova,Vera Kokhanova
+vputinski,Victoria Putinski
+vjirinec,Vitek Jirinec
+wkennerley,Will Kennerley
+wenglish,Willow English
+yhhsu,Yu-Hsun Hsu (Echo)
+zharris,Zack Harris
+zlebrun-southcott,Zoe Lebrun-Southcott
+cmontgomerie,Claire Montgomerie
+emagnuson,Emily Magnuson
+mcorrell,Maureen Correll

data/schema-build-script.sql

@@ -0,0 +1,76 @@
+.nullvalue -NULL-
+
+CREATE TABLE Species (
+    Code VARCHAR PRIMARY KEY,
+    Common_name VARCHAR UNIQUE NOT NULL,
+    Scientific_name VARCHAR, -- can't make NOT NULL, missing data in some rows
+    Relevance VARCHAR
+);
+COPY Species FROM 'species.csv' (header TRUE);
+
+CREATE TABLE Site (
+    Code VARCHAR PRIMARY KEY,
+    Site_name VARCHAR UNIQUE NOT NULL,
+    Location VARCHAR NOT NULL,
+    Latitude FLOAT NOT NULL CHECK (Latitude BETWEEN -90 AND 90),
+    Longitude FLOAT NOT NULL CHECK (Longitude BETWEEN -180 AND 180),
+    Area FLOAT NOT NULL CHECK (Area > 0),
+    UNIQUE (Latitude, Longitude)
+);
+COPY Site FROM 'site.csv' (header TRUE);
+
+CREATE TABLE Personnel (
+    Abbreviation VARCHAR PRIMARY KEY,
+    Name VARCHAR UNIQUE NOT NULL
+);
+COPY Personnel FROM 'personnel.csv' (header TRUE);
+
+CREATE TABLE Camp_assignment (
+    Year INTEGER NOT NULL CHECK (Year BETWEEN 1950 AND 2015),
+    Site VARCHAR NOT NULL,
+    Observer VARCHAR NOT NULL,
+    Start DATE,
+    "End" DATE,
+    FOREIGN KEY (Site) REFERENCES Site (Code),
+    FOREIGN KEY (Observer) REFERENCES Personnel (Abbreviation),
+    CHECK (Start <= "End"),
+    CHECK (Start BETWEEN (Year||'-01-01')::DATE AND (Year||'-12-31')::DATE),
+    CHECK ("End" BETWEEN (Year||'-01-01')::DATE AND (Year||'-12-31')::DATE)
+);
+COPY Camp_assignment FROM 'ASDN_Camp_assignment.csv' (header TRUE);
+
+CREATE TABLE Bird_nests (
+    Book_page VARCHAR,
+    Year INTEGER NOT NULL CHECK (Year BETWEEN 1950 AND 2015),
+    Site VARCHAR NOT NULL,
+    Nest_ID VARCHAR PRIMARY KEY,
+    Species VARCHAR NOT NULL,
+    Observer VARCHAR,
+    Date_found DATE NOT NULL
+        CHECK (
+            Date_found BETWEEN (Year||'-01-01')::DATE
+            AND (Year||'-12-31')::DATE
+        ),
+    how_found VARCHAR CHECK (how_found IN ('searcher', 'rope', 'bander')),
+    Clutch_max INTEGER CHECK (Clutch_max BETWEEN 0 AND 20),
+    floatAge FLOAT CHECK (floatAge BETWEEN 0 AND 30),
+    ageMethod VARCHAR CHECK (ageMethod IN ('float', 'lay', 'hatch')),
+    FOREIGN KEY (Site) REFERENCES Site (Code),
+    FOREIGN KEY (Species) REFERENCES Species (Code),
+    FOREIGN KEY (Observer) REFERENCES Personnel (Abbreviation)
+);
+COPY Bird_nests FROM 'ASDN_Bird_nests.csv' (header TRUE);
+
+CREATE TABLE Bird_eggs (
+    Book_page VARCHAR,
+    Year INTEGER NOT NULL CHECK (Year BETWEEN 1950 AND 2015),
+    Site VARCHAR NOT NULL,
+    Nest_ID VARCHAR NOT NULL,
+    Egg_num INTEGER NOT NULL CHECK (Egg_num BETWEEN 1 AND 20),
+    Length FLOAT NOT NULL CHECK (Length > 0 AND Length < 100),
+    Width FLOAT NOT NULL CHECK (Width > 0 AND Width < 100),
+    PRIMARY KEY (Nest_ID, Egg_num),
+    FOREIGN KEY (Site) REFERENCES Site (Code),
+    FOREIGN KEY (Nest_ID) REFERENCES Bird_nests (Nest_ID)
+);
+COPY Bird_eggs FROM 'ASDN_Bird_eggs.csv' (header TRUE);

data/site.csv

@@ -0,0 +1,17 @@
+Code,Site name,Location,Latitude,Longitude,Total_Study_Plot_Area_(ha)
+barr,Barrow,"Alaska, USA",71.3,-156.6,220.4
+burn,Burntpoint Creek,"Ontario, Canada",55.2,-84.3,63.0
+bylo,Bylot Island,"Nunavut, Canada",73.2,-80.0,723.6
+cakr,Cape Krusenstern,"Alaska, USA",67.1,-163.5,54.1
+cari,Canning River Delta,"Alaska, USA",70.1,-145.8,722.0
+chau,Chaun River Delta,"Chukotka, Russia",68.8,170.6,248.2
+chur,Churchill,"Manitoba, Canada",58.7,-93.8,866.9
+coat,Coats Island,"Nunavut, Canada",62.9,-82.5,1239.1
+colv,Colville River Delta,"Alaska, USA",70.4,-150.7,324.8
+eaba,East Bay,"Nunavut, Canada",64.0,-81.7,1205.5
+iglo,Igloolik,"Nunavut, Canada",69.4,-81.6,59.8
+ikpi,Ikpikpuk,"Alaska, USA",70.6,-154.7,174.1
+lkri,Lower Khatanga River,"Krasnoyarsk, Russia",72.9,106.1,270.9
+made,Mackenzie River Delta,"Northwest Territories, Canada",69.4,-135.0,667.3
+nome,Nome,"Alaska, USA",64.4,-164.9,90.1
+prba,Prudhoe Bay,"Alaska, USA",70.3,-148.6,120.0